Functional Erythropoietin Receptor of the Cells with Neural Characteristics COMPARISON WITH RECEPTOR PROPERTIES OF ERYTHROID CELLS*

Radioiodinated erythropoietin (Epo) was bound spe- cifically to the cells of two non-erythroid clonal lines, PC12 and SN6, which expressed neuronal characteristics. The binding was time-, cell number-, and dose- dependent and was reversible. Although the cloned Epo receptor from PC12 cells (derived from rat adrenal medulla) was identical to that from rat erythroid cells, significant differences in the ligand binding properties between two cell lineages were found; 1) PC12 cells had a single class of binding sites with very low affinity (& = 16 nM), whereas erythroid cells had two classes of binding sites with different affinities (Kd = 95 PM for high affinity sites and 1.9 nM for low affinity sites), and 2) cross-linking experiments revealed one cross-linked product of 105 kDa for PC12 cells and two products of 140 and 120 kDa for erythroid cells. Taken together with additional results, the presence of a pu- tative accessory protein(s) that may alter the ligand binding affinity through interaction with Epo receptor is discussed. The binding of Epo to PC12 cells caused a rapid increase in the HuEpo was pro- duced (14) and isolated (15) as described previously. The fully deglycosylated rHuEpo was prepared by the procedures also described previously (16). Rat Epo was isolated from the sera of anemic ani-mals.' Quantities of rHuEpo and rat Epo were measured by sandwich- type enzyme-linked immunoassay (15, 17). Recombinant HuEpo was radioiodinated at a specific activity of 1.11 MBq/pg protein (18). Experiments for the binding of '251-rHuEpo to cells were done as described previously (19). Binding assay mixtures contained cells, 20 mM HEPES, pH 7.4, 0.1% bovine serum albumin, 0.1% NaN3 (an inhibitor of the ligand internalization), and rHuEpo in a total volume of 150 pl PBS. Incubations for binding were done at 15 "C for 3 h unless otherwise indicated. The cells were pelleted, washed once with PBS, and suspended in 200 pl of PBS. The suspension was layered on an 800-111 cushion buffer (PBS containing 10% bovine serum albumin), and the cells were separated from the unbound ligand by centrifugation. The tube contents were frozen in solid C02/ethanol. The tip was cut off just above the cell pellet, and the radioactivity was Specific binding was as the difference in bound radioactivity between samples incubated in the absence and the presence of 200-fold unlabeled Epo.

The action of Epo on erythroid precursor cells has been only a generally accepted function; Epo supports survival of the cells and stimulates their proliferation and differentiation (see Ref Epo-R is present in the cells from rodent placentas (5). A physiological significance of these findings remains to be proven but such findings of non-erythroid cells bearing Epo receptors may provide us with an opportunity to find a new physiological function of Epo.
Cells derived from the neural crest lineage appear to display in culture either neuronal or chromaffin phenotypes. NGF induces neuronal characteristics and corticosteroids potentiate chromaffin properties (6). The rat cell line PC12, which has been established from an adrenal medullary pheochromocytoma, has similar bipotent properties; the cells express more neuronal properties with exposure to NGF and keep chromaffin properties with corticosteroids (7)(8)(9). T h e cell line SN6, a clonal hybrid cell line developed from the septal region of the mouse basal forebrain, expresses characteristics typical of cholinergic neurons (10). Here we report the presence of Epo-R in these two cell lines, molecular properties of Epo-R on PC12 cells compared with that on erythroid cells, and the Epo-induced increase in intracellular concentrations of calcium and monoamines of PC12 cells. The presence of Epo-R on these neural cell lines can be rationalized on the basis of recent findings that Epo augments choline acetyltransferase activity in primary cultured neurons and supports the in vivo survival of lesioned neurons (ll), although the physiological significance of these findings has not been verified. Comparison of Epo-R on the neural cells with that on erythroid cells, therefore, is important for studying a new physiological function of Epo and also for understanding the Epo-induced signal transduction pathway, including identification of protein(s) on neural cells involved in the interaction with Epo. 3, Collaborative Research, Inc.; human IL-6; tumor necrosis factor and GM-CSF, Genzyme; human EDF (activin A), a kind gift from Dr. Eto of Ajinomoto; human HGF, a kind gift from Dr. Higashio of Snow Brand Milk Products; mouse NGF, Biochemical Technologies Inc.; bovine insulin, Sigma; Oligo(dT)-Latex, Takara.
Cells and Cell Culture-PC12 pheochromocytoma cells were cultured in DMEM supplemented with 10% horse serum and 5% FCS under a humidified atmosphere containing 10% Con, and SN6.10.2.2 cells (a kind gift from Dr. B. H. Wainer) were cultured in DMEM supplemented with 10% FCS in the presence of 5% COz. The latter is a subclone of the SN6 cell line (10). PC12 and SN6 cells were removed from the flasks and dissociated by pipetting to yield single cell suspensions. The spleen single cell suspension enriched with Eporesponsive cells was prepared from Wistar rats made anemic by injections of phenylhydrazine (12). Chromaffin cells were prepared from rat adrenal medullas (13). Adrenal glands were removed from 6-week-old Wistar rats. Medullas were carefully freed from capsular and cortical tissue. The medullary tissue was cut into small pieces and kept in Ca'+-free Hanks' balanced salt solution at 37 "C for 20 min. The dissected medullas were transferred to MEM containing 10% FCS, 0.15% collagenase, 30 pg/ml DNase I, 10 mM HEPES and digested at 37 "C for 100 min with constant shaking. Every 20 min during digestion, the digestion medium was renewed. The cells were washed three times with MEM containing 10% FCS by centrifugation at 50 X g for 3 min. More than 90% of the cells were viable judging from staining with trypan blue.
Epo and Its Binding to the Cells-Recombinant HuEpo was produced (14) and isolated (15) as described previously. The fully deglycosylated rHuEpo was prepared by the procedures also described previously (16). Rat Epo was isolated from the sera of anemic animals.' Quantities of rHuEpo and rat Epo were measured by sandwichtype enzyme-linked immunoassay (15,17). Recombinant HuEpo was radioiodinated at a specific activity of 1.11 MBq/pg protein (18). Experiments for the binding of '251-rHuEpo to cells were done as described previously (19). Binding assay mixtures contained cells, 20 mM HEPES, pH 7.4, 0.1% bovine serum albumin, 0.1% NaN3 (an inhibitor of the ligand internalization), and rHuEpo in a total volume of 150 pl PBS. Incubations for binding were done at 15 "C for 3 h unless otherwise indicated. The cells were pelleted, washed once with PBS, and suspended in 200 pl of PBS. The suspension was layered on an 800-111 cushion buffer (PBS containing 10% bovine serum albumin), and the cells were separated from the unbound ligand by centrifugation. The tube contents were frozen in solid C02/ethanol. The tip was cut off just above the cell pellet, and the radioactivity was counted. Specific binding was defined as the difference in bound radioactivity between samples incubated in the absence and the presence of 200-fold unlabeled Epo. Scatchard plot analyses of the binding data were performed by the LIGAND program (20).
Internalization of Epo Bound to PC 12 Cells-Internalization of 1251-rHuEpo bound to PC 12 cells was measured by the low-pH method (21). Binding assay mixtures containing 5.6 X lo6 cells, 7.5 nM "' I-rHuEpo, and no NaN3 were incubated at 15 "C for 3 h or at 37 "C for 1 h. The cell-associated ligand at 15 "C is equivalent to the cell surface receptor-bound ligand, whereas that at 37 "C contains both the receptor-bound ligand and the internalized ligand. Cells were washed three times with PBS and then suspended in ice-cold 0.25 M acetic acid, pH 2.5, containing 0.5 M NaCl or in PBS. The receptor-bound Epo should be released upon incubation at the acidic pH, but the internalized Epo should remain cell-associated. The cell suspensions were incubated at 0 "C for 3 min and layered on the cushion buffer. After centrifugation, the tube contents were frozen, and the tips were cut off as described previously. Radioactivity of the tips represents the internalized Epo and that of the supernatant represents Epo released from Epo-receptor complexes on the cell surface. Control runs were done in the presence of 200-fold unlabeled Epo.
Cross-linked Products-Cross-linked products between Iz5I-rHuEpo and EPO-R were prepared as described previously (18).
Anti R lacking cytoplasmic and transmembrane domains was produced and isolated (22). Rabbit anti-mouse soluble Epo-R antiserum was  prepared by injection of the isolated soluble Epo-R. Rabbit anti-NH2terminal mouse Epo-R antiserum was prepared using the 15 NHZterminal amino acid peptide conjugated to keyhole limpet hemocyanin . .
.~ PreDaration of rHuEDo-fixed Gel-For preparation of the Epo-fixed gel, 15'0 mg of rHuEpo'in '100 mM NaHCOawas gently mixed with 9 ml of CH-Sepharose 4B gel at 20 "C for 1 h and at 4 "C overnight. The gel was pelleted by centrifugation and then suspended in 100 mM Tris-HC1, pH 8.0, to block the remaining sites on the gel. Washings ofthe gel were done four times each with the acidic solution (50 mM acetic acid containing 500 mM NaCI) and then with the basic solution (50 mM Tris containing 500 mM NaC1). The Epo-fixed gel was kept in PBS containing 0.1% NaN3 at 4 "C before use.
Immunological Detection of Epo-R Solubilized from PC12 Cells and Rat Erythroid Cells by Western Blotting-Epo-R solubilized from PC12 cells and rat spleen erythroid cells was concentrated using the Epo-fixed gel and then identified with the Western blotting technique. About 10' cells were lysed by incubation at 4 "C for 1 h in 2 ml buffer A, PBS containing 0.5% (w/v) CHAPS, 10 p~ APMSF, 10 p~ leupeptin, and 1 mM EGTA. The lysate was centrifuged 12,000 X g for 30 min at 4 "C. The supernatant was mixed with 15 pl of rHuEpofixed gel (15 mg of rHuEpo/ml of gel) overnight at 4 "C. The gel was pelleted by centrifugation and washed three times with 500 p1 of buffer A. Proteins bound to the Epo-fixed gel were solubilized in 50 p1 of SDS-buffer consisting of 60 mM Tris-HC1, pH 6.8, 2% SDS, 10% glycerol, 1.4 M 2-mercaptoethanol, and 0.001% bromphenol blue. Solubilized proteins were separated by electrophoresis with SDS, 9% polyacrylamide gel. Western blotting was carried out according to the method of Burnette (23) with some modifications. Briefly, polyacrylamide gel was immersed in 50 ml of transfer buffer consisting of 48 mM Tris, 39 mM glycine, 1.3 mM SDS, and 20% methanol for 15 min, with two changes in buffer solution. The proteins in gel were transferred to a 0.45-pm nitrocellulose filter at 1.2 mA/cm2 for 40 min. The nitrocellulose filter was immersed in 15 ml of block ace at 4 'C overnight for blocking. The filter was then dipped into 15 ml of buffer B consisting of 0.05% Tween 20, 5% block ace in PBS, and the antisoluble Epo-R antiserum. After the incubation for 1 h at room temperature, the filter was washed three times with buffer B and then immersed in 15 ml of buffer B containing peroxidase-fixed goat anti-rabbit IgG (1 pg/ml) for 1 h at room temperature and then washed five times with buffer B. The antigen, Epo-R, was visualized using the enhanced chemiluminescence Western blotting detection system.
Northern Analysis of Epo-R mRNA from PC 12 Cells and Erythroid Cells-Total RNA from PC12 cells was prepared using the method for preparation of cytoplasmic RNA (24). Total RNA from anemic rat spleen and fetal mouse liver cells was prepared according to the manufacturer's instruction (Pharmacia) using guanidinium isothiocyanate, CsCI, and trifluoroacetic acid. Poly(A)+ RNA was isolated using oligo(dT)-Latex. Poly(A)+ RNA (8 pg) treated with glyoxal/ dimethyl sulfoxide (24) was loaded in each lane of a 1% agarose gel. After electrophoresis, the RNA was blotted onto Hybond N filter, and the filter was hybridized with the 32P-labeled coding region of rat Epo-R cDNA between the F and R primers (see below) as a probe.
Nucleotide Sequence of Epo-R cDNA from PC12 Cells and Rat Erythroid Cells-The coding regions of Epo-R cDNAs from anemic rat spleen cells and PC12 cells were obtained using the RT-PCR method (25). Primers N f5'-GGCAAGCTTGGGCTCCATCATG-3')~ C (5' -GCTCTAGAGTAGGCTGGAGTCC' ),TGI AGCAACCTGCG-3'), and R (5"GTCCAGGATGGTGTACTCA") were synthesized to amplify the coding region of Epo-R from PC12 performed between primers N and C, N and R, or F and C using single-stranded cDNAs as templates. Nucleotide sequences of RT-PCR products were determined by the direct sequencing method (27) and those of cloned RT-PCR products by the dideoxy sequencing method using Sequenase (28). Calcium Concentration-Intracellular calcium concentrations were determined using the fluorescent calcium indicator fura-2 (29,30). PC12 cells were loaded with 10 p~ fura-BAM, the ester form of fura-2, by incubating the cell suspensions in PBS at 2 X IO6 cells/ml for 45 min at 37 "C. The cells were washed three times by an isotonic buffer consisting of 125 mM NaCl, 5 mM KCl, 1.2 mM MgC12, 1.2 mM KHzP04, 5 mM NaHC03, 1 mM CaCl2, 6 mM glucose, 25 mM HEPES, and 1 mM CaCIZ and resuspended in the same buffer. One-milliliter aliquots of the cell suspension were pipetted into tubes and the tubes were kept on ice for 1 h. Cells were resuspended in the same buffer warmed at 37 "C. Fluorescence measurements were made in 1-ml samples continuously stirred in quartz-glass cuvettes and thermostatically maintained at 37 "C. Fluorescence was monitored with a Shimazu RF-5000 spectrofluorimeter, with excitation at dual wavelengths (340 and 380 nm) and emission at 490 nm; ratios of fluorescence intensities emitted when excited at 340 and 380 nm were recorded.
Intracellular Monoamine Concentrations of PC 12 Cells-PC12 cells (1.2 X IO6 cells/3 ml of medium) were cultured in a 35-mm plastic dish for 2 days in the presence of Epo at 3 nM or in its absence. Cells were thoroughly washed with PBS and suspended in 100 pl PBS. An aliquot (20 pl) of the suspension was used for measurement of protein according to Lowry's method after the lysis of cells in 1% SDS. The remainder (80 rl) of the cell suspension was homogenized with 200 pl of 0.1 M perchloric acid. After centrifugation, the supernatant was passed through a 0.22-pm filter. Monoamine contents in the filtrate were measured with ESA's Coulometric-detector Array Gradient System (Neurochem Nikko Bioscience). We examined whether the binding of Epo to PC12 cells was reversible. Radioiodinated rHuEpo was bound to cells by incubating the binding mixture at 15 "C for 3 h. The cells were washed thoroughly to remove the free ligand. The washed cells were again incubated at 15 "C, and the decrease in the bound ligand during incubation was measured. Fig. 2 shows the time-dependent release of radioactivity from PC12 cells and rat erythroid cells. The dissociated radioactivity was precipitated almost completely by 5% trichloroacetic acid and migrated in SDS-polyacrylamide gel with a molecular size similar to that of '*'I-rHuEpo; binding of Epo to PC12 cells as well as to erythroid cells is reversible. Epo appears to dissociate more rapidly from PC12 cells than erythroid cells.

Binding
Since the unlabeled Epo is not added in the ligand-dissociation mixture, however, the time-dependent release of Epo in Fig. 2 reflects both dissociation of Epo from the cells and reassociation of the dissociated ligand. Table I shows that Epo bound to PC12 cells was internalized at 37 "C but not at 15 "C. After the cells were incubated with '251-rH~Epo at 15 or 37 "C, the cell-associated Epo was exposed to a neutral pH or an acidic pH and then centrifuged. At the acidic pH the ligand bound to the cell surface receptor would be released from the cells, and therefore the radioactivity should appear in the supernatant after centrifugation. But the internalized ligand would not be released from the cells upon low-pH treatment, and therefore, the radioactivity C FIG. 1. Binding of '''I-rHuEpo to PC12, SN6, and rat spleen erythroid cells. A , time dependence of binding to PC12 cells. Binding mixtures contained 7.5 nM '"1-rHuEpo and 4.5 X lo6 cells. 0, total binding; 0, nonspecific binding, and M, specific binding. B, cell number-dependency of binding to PC12 cells. 0, total binding; 0, nonspecific binding; specific binding. Binding mixtures contained 5 nM Iz5I-rHuEpo. C, Scatchard plots of ligand-saturation curves of PC12 cells ( 0 ) and SN6 cells  should be associated with the pelleted cells. When Epo was bound to the cells at 15 "C and then the cell-associated Epo was exposed to the neutral pH, most of the radioactivity was associated with the pelleted cells. The radioactivity, however,

FIG. 2. Reversible binding of Epo to PC12 cells and rat erythroid cells.
Binding mixtures were as described under "Experimental Procedures" except that the mixtures of rat spleen erythroid cells contained 0.5 nM Iz5I-rHuEpo and 4 X lo6 cells, and the mixtures of PC12 cells contained 5 nM '251-rHuEpo and 2.5 X lo6 cells. The mixtures were incubated at 15 "C for 3 h for binding of Epo and then the cells were washed three times by PBS. The washed cells were suspended in 150 p1 of PBS. The suspensions were incubated at 15 "C for indicated periods and centrifuged on the cushion buffer. Radioactivity of the cell-associated Epo was counted. The total cell-associated radioactivity of rat spleen cells at zero time was 4454 f 183 cpm (mean of duplicate assays f deviation of duplicate values), and the nonspecific binding was 1194 f 37 cpm; therefore, specific binding was 3260 cpm. The total and nonspecific binding of PC12 cells was 6536 f 289 and 3763 + 176 cpm, respectively; specific binding was 2773 cpm. Specific binding was defined as 100%. Nonspecific binding at each point was measured using the cells bound to '2sII-Epo in the presence of 200-fold unlabeled Epo in the binding mixtures. 0, rat erythroid cells; 0, PC12 cells. The supernatants that contained the radioactivity released from the cells were used for identification of the released radioactive material (see text). was almost completely removed from the cells when the cellassociated Epo was exposed at the acidic pH. When the binding was done at 37 "C, approximately 40% of the total specific binding was still cell-associated after exposure to the acidic pH, in agreement with erythroid cells (21). The specific binding of Epo to PC12 cells a t 15 "C is indeed equivalent to the receptor-bound ligands on the cell surface, but a significant portion of the ligands bound to the cells a t 37 "C is internalized.
The culture of PC12 cells in the presence of NGF caused neurite growth. When Epo binding was tested for PC12 cells cultured for 3 days in the presence of 2 nM NGF, there were no significant changes in the characteristics of Epo binding, including the number of binding sites.
PC12 cells were derived from rat adrenal chromaffin cells (6). The natural chromaffin cells were prepared from rat adrenal medullas, and the specific binding of lZ5I-rHuEpo a t 10 nM was tested using 4 X lo5 cells. There was specific and reproducible binding (410 & 20 cpm in triplicate assays), but low numbers of the available cells did not allow us to carry out further characterization of the binding.
Specificity of Binding of Epo to PC12 Cells-Affinities of Epo binding sites on PC12 and SN6 cells are much lower than those on erythroid cells, and Epo is a heavily glycosylated protein (1, 16). It is possible that the Epo binding to these neuronal cells occurs through recognition of sugar chains attached to Epo, like a lectin-glycoprotein interaction. This possibility, however, was excluded by the results that the deglycosylated form of rHuEpo inhibited the specific binding of '251-rHuEpo with a potency similar to that of the fully glycosylated rHuEpo and that metal chelators (EGTA and EDTA) and monosaccharides (GalNAc and Gal), which are inhibitors of the interaction among some lectins and glycoproteins, showed no effect on the specific binding (Table 11). Rat Epo also inhibited the specific binding of '"I-rHuEpo with an efficiency similar to that of the unlabeled rHuEpo, indicating that the low affinity of PC12 cells was not due to a combination heterogeneous in origin of the ligand and target cells. The monoclonal antibodies (R2 and R6) against rHuEpo decreased the binding, probably by deprivation of free ligand, but the control antibody directed against transglutaminase   Cross-linked Products-Epo receptors on PC12 cells and rat erythroid cells were affinity-labeled by a chemical crosslinker, solubilized with Triton X-100, and then analyzed by SDS-polyacrylamide gel electrophoresis under reducing conditions (Fig. 3). There were two cross-linked products of rat erythroid cells with the molecular masses of 120 and 140 kDa, and a single product of PC12 cells with a 105-kDa molecular mass. These products were not detected when an excess of unlabeled Epo was added to the binding mixture. Subtracting the molecular mass of Epo, 35 kDa, from those of the cross-linked products gave the apparent size of the component cross-linked to Epo in each product; 85 and 105 kDa of erythroid cells and 70 kDa of PC12 cells.

Immunochemical Detection of Epo-R of PC12 Cells and
Erythroid Cells-Rabbit anti-mouse Epo-R antiserum was prepared by using the soluble form of mouse Epo-R as an antigen. This antiserum was used to identify Epo-R of erythroid cells and PC12 cells. Epo-R was solubilized from cells and concentrated by the Epo-fixed gel. The concentrated Epo-R was detected by the Western blotting technique (Fig. 4). Epo-R of mouse erythroid cell line, TSA8, was detected as a single band with a molecular size of 68 kDa (lane 1 ). A band with a similar size was found for rat spleen cells (lane 2). These bands were undetectable when the antiserum was preadsorbed with the antigen, soluble form of mouse Epo-R (lanes 4 and 5). PC12 cells yielded one major band a t 62 kDa, and two additional minor bands a t 58 and 54 kDa (lane 3 ) . The major 62-kDa band appears to be Epo-R on PC12 cells, because the minor bands were still detected by the preadsorbed antiserum (lane 6). Experiments using the anti-NH2-terminal antiserum also showed Epo-R of 68 kDa in TSA8 and 62 kDa in PC12 cells.
Northern Blotting of Epo-R mRNA-Poly(A)+ RNA was isolated from fetal mouse liver cells, rat spleen cells, and PC12 cells and subjected to Northern hybridization using rat Epo-R cDNA as a probe (Fig. 5). The  Nucleotide Sequence of Epo-R of PC12 Cells and Rat Erythroid Cells-The affinity of Epo to PC12 Epo-R is very low as compared with that of erythroid cells (Fig. l), and the size of PC12 Epo-R is smaller than that of erythroid cells (Fig. 4). In order to know whether this low affinity is due to expression of a deletion-mutant Epo-R on PC12 cells, we determined the nucleotide sequence of the entire coding region of Epo-R cDNA from rat erythroid cells and PC12 cells. The nucleotide sequence of PC12 cells was identical to that of rat erythroid cells. Fig. 6 shows the nucleotide sequence of rat Epo-R and its deduced amino acid sequence; the matured rat Epo-R consists of 483 amino acids, and its calculated molecular weight is 52,794. For comparison, the amino acid sequences of mouse (26) and human (32, 33) Epo-R are shown. The amino acid sequences of the matured proteins were 82% conserved between the rat and human Epo-Rs.
Homology between rat and mouse increased to 94%. Insertion of one amino acid occurs a t position 49 in human Epo-R. The

FIG. 4. Immunochemical identification of Epo-R of PC12 cells and erythroid cells. Rat spleen erythroid cells, PC12 cells. and mouse erythroleukemia cells (TSAR) were solubilized by CHAPS.
Epo-R in the lysates was concentrated using the Epo-fixed gel. The concentrated Epo was detected by the Western blotting technique using anti-soluble Epo-R antiserum. In A, the anti-soluble Epo-R antiserum was used without the antigen pread.sorption, whereas. in R, the antiserum was preincubated with excess antigen. soluble Epo-R. Lanes I and 4, TSA8 cells; lanes 2 and .5, rat spleen cells: lanes 3  and 6, PC12 cells.  (34,35). The WS motif of Epo-R appears to be critical for protein folding, ligand binding, and signal transduction (36,37). One

N -
glycosylation site (NYS a t positions 51-53) is conserved in rodent and human Epo-Rs. In the cytoplasmic domain, there is a well conserved region (positions 248-295) proximal to the transmembrane domain. This region is homologous to the IL-2 receptor chain (34) and contains the sequence that appears to play a key role in expression of the growth signal including tyrosine kinase activity (38)(39)(40). Forty amino acid residues in the carboxyl terminus have been proposed as a negative regulatory region (40). This regulatory region is preceded with a highly conserved region (positions 418-446) whose significance is not known.
Effects of Epo on Cultured PC12 Cells-There were no effects on ["]thymidine incorporation into DNA of PC12 cells. Incubation of PC12 cells with Epo caused a rapid and transient increase in the cytosolic free calcium concentration (Fig. 7A ). This increase was completely abolished by the Epodirected monoclonal antibodies R2 and R6 (not shown). An increase in the calcium concentration also occurred with the addition of bradykinin, a control compound that has been known (30) to mobilize calcium from intracellular stores to the cytoplasmic fraction (Fig. 7A). EGTA abolished the increase in calcium by Epo but had no effect on the bradykinininduced increase (Fig. 78). Epo-induced increase in the cytosolic concentration was dose-dependent; when Epo was added a t 30 PM, 900 PM, 3 nM, and 6 nM, increases of the calcium concentration in percent were 10 f 6.3 ( n = lo), 49 f 13.9 ( n = 12, p < 0.025), 60 f 13.9 ( n = 12, p < 0.025), and 70 f 17.0 ( n = 7, p < 0.025), respectively, where the calcium concentration without addition of Epo was defined as 100 and the increased values were means f S.E. Addition of Epo a t 9 nM resulted in a rather low increase of 36 f 11.6 ( n = 12, p < 0.025) by an unknown reason.
A culture of PC12 cells in the presence of Epo elevated intracellular concentrations of monoamines such as DOPA, dopamine, DOPAC, and HVA; MDOPA remained unchanged (Fig. 8). This may suggest that tyrosine hydroxylase, a key enzyme in the biosynthetic pathway of the four monoamines that increased, is activated or accumulated.

DISCUSSION
Stimulation of erythropoiesis has been thought to be an exclusive physiological function of Epo. Findings that Epo-R exists on megakaryocytes (2), endothelial cells from human umbilical vein and bovine adrenal capillary (3), R-lymphocytes (4), rodent placenta cells (51, multipotential hematopoietic stem cells (42), and embryonic stem cells (42. 43), however, may suggest yet unidentified functions of Epo, although some of these findings may not be physiologically meaningful. The PC12 pheochromocytoma cell line has been a widely used model system for studying the action mechanism of NGF on neurons, because this cell line undergoes neurite outgrowth responding to NGF (7). SN6 cells display some properties of cholinergic neurons such as high choline acetyltransferase activity and neurite extension (8). In this paper we described that these cell lines had Epo-binding proteins. Detailed analyses of PC12 cells and rat erythroid cells, including nucleotide sequence analyses of Epo-R, revealed that the Epo-binding protein on PC12 cells is the counterpart of erythroid cells. Expression of Epo-R on these neural cell lines seems not to be accidental, because Epo augments choline acetyltransferase activity in mouse embryonic primary septal neurons and supports survival of septal cholinergic neurons in adult rats (11). Epo exerts i t s activity on these neurons a t a nanomolar range ( l l ) , which is comparable with Kd of Epo-R on PC12 and SN6 cells. A prerequisite for inference of a physiological significance of Epo-R in nerve cells is to demonstrate production of Epo in brain. Currently we are attempting to find Epo production in brain tissue.
Immunochemical detection of solubilized Epo-R showed that the size (62 kDa) of Epo-R from PC12 cells was smaller than that (68 kDa) from rat erythroid cells. Analyses of PC12 mRNA by Northern blotting and its nucleotide sequence, however, were indicative of expressing neither a mutated Epo-R nor an alternative splicing-derived Epo-R. Neural cells may differ from erythroid cells in post-translational processing of Epo-R, resulting in expression of Epo-R with different sizes. One putative N-glycosylation site existing in the extracellular domain of Epo-R could be a cause for the size difference, but the difference could also result from other processings such as proteolysis and phosphorylation. Proteolytic removal of the NH2-terminal region from PC12 Epo-R is unlikely, because the antiserum against the NH,-terminal peptide reacts with the solubilized Epo-R.
The ligand affinity of Epo-R on neural cells (K,, = 10 -16 nM) is significantly lower than those on erythroid cells ( K d = 95 PM for high affinity site and 1.9 nM for low affinity site). The low affinity of Epo-R on PC12 cells might be related to a post-translational processing that yields Epo-R with a smaller size. But the N-linked sugar, if it causes the difference in Epo-R size, is not responsible for the affinity difference, because the N-glycosylation site-defective mutant of mouse Epo-R (44) and its extracellular soluble domain (22) are similar to the respective wild-type counterpart in binding with Epo.
A more intriguing hypothesis to account for the affinity difference is that there are accessory proteins that interact with Epo-R, altering interaction of Epo-R with the ligand. Cross-linking experiments revealed the presence of two proteins with 105 and 85 kDa in erythroid cells and a 70-kDa protein in PC12 cells (the size of Epo, 35 kDa, has been subtracted from the cross-linked products) (see Fig. 3). The two proteins found in rat erythroid cells are consistent with  second, third, and fourth lines, respectively. Position 1 was given to the putative NH2-terminal amino acid of the matured protein (26). The transmembrane domain is underlined. Amino acid sequences of mouse (26) and human (32,33) were cited from the indicated sources. The nucleotide sequence data reported in this paper will appear in the DDBJ, EMBL, and GenBank Nucleotide Sequence Databases with accession number D13566. those reported previously in rat source (45) and the presence of two components with similar sizes has been shown in erythroid cells from human and mouse (for review see Refs. 1 and 46). No conclusive evidence for these proteins being derived from cloned Epo-R cDNA has been reported. The molecular weight of Epo-R predicted from cloned mouse (26) and rat cDNA (this paper) is about 53,000, and the size of Epo-R detected by immunoblotting (Ref. 47 and this paper) or by ligand blotting technique (48,49) is around 65 kDa; this size of Epo-R is much smaller than two proteins included in cross-linked products of erythroid cells. From these, it has been suspected that erythroid cross-linked products may consist of Epo, Epo-R, and an unidentified accessory protein. It has been reported recently, however, that erythroid crosslinked products before denaturation are precipitated by the antiserum against the cytoplasmic domain of mouse Epo-R, but, surprisingly, they are not precipitated upon denaturation (boiling of the cross-linked products for 5 min in Laemmli SDS-electrophoresis buffer) (50). The cross-linked products either before or after denaturation could be precipitated by the antiserum against Epo. These results have been interpreted as a strong indication of the following. 1) Epo interacts with Epo-R, but the reagent used does not efficiently crosslink between Epo and Epo-R 2 ) Epo is cross-linked with accessory protein(s) which is immunochemically unrelated to Epo-R, and 3) before denaturation of cross-linked products, Epo-R is retained in complexes through noncovalent interaction with Epo or both Epo and an accessory protein, but Epo-R is dissociated from the complexes upon denaturation. By using the antiserum against the extracellular domain of mouse Epo-R, we performed similar experiments of erythroid cross-linked products, and the data obtained were consistent with those already reported (50). The 105-kDa cross-linked product of PC12 cells could not be precipitated either before or after denaturation,3 although our antiserum reacted with the solubilized Epo-R of PC12 cells on the Western blotting filter (see Fig. 4). The PC12-derived cross-linked product was precipitated by the antiserum against Epo, regardless of denaturation of the product. From these results of the PC12 cells, we infer that. most of the Epo-R associated with the PC12 cross-linked product may have dissociated during solubilization of the cross-linked product so that the cross-linked product even before being subjected to the denaturation treat--,'S. Masuda, M. Nagao, and R. Sasaki, unpublished data. ment can not be precipitated by the Epo-R antiserum. This inference is supported by rapid dissociation of Epo .Epo-R complexes on PC12 cells (see Fig. 2). Based on such information, our hypothesis is that the cloned Epo-R is a common molecule required for Epo binding, but the ligand affinity is modulated by multiple accessory proteins of which expression may be cell type-specific. Affinity of the Epo-R on endothelial cells is also very low (Kd = 2-8 nM), and the size of a major cross-linked product is around 79 kDa (4). Homodimerization of the cloned Epo-R has been implicated in the signal transduction pathway of Epo (51), but this does not exclude the possible involvement of a heterosubunit in the pathway. For high affinity receptor sites of IL-2 ( X ) , IL-3 (53), IL-5 (54), IL-6 (55), and GM-CSF (53), oligomerization of heterosubunits is required. Epo caused a rapid increase in calcium in PC12 cells. EDTA did not interfere with binding of Epo to the cells (Table 11) but inhibited the Epo-induced increase of calcium, indicating that Epo stimulates mostly the calcium influx from outside of PC12 cells. The role of calcium in Epo-signal transduction of erythroid cells has been controversial (56-60). Several studies have indicated significant alterations in intracellular free calcium concentration or calcium influx in response to Epo (56-59). Experiments with purified Epo-responsive mouse erythroid cells that probably express Epo-R at the highest number, however, showed that the Epo action was not accompanied by an acute alteration in intracellular calcium concentration (60). It has been pointed out that the conflicting data may result from different maturation stages of the cells used (60). The data may also be dependent on whether or not the cells have been exposed in vivo to a high Epo concentration before the cells were prepared from animals. Further studies are necessary to find a definitive role of calcium in Epo-signal transduction.