Identification of the Prostacyclin Receptor by Radiation Inactivation*

Evidence has been obtained for a specific protein receptor for prostacyclin on cells of the NCB-PO SO-matic hybrid. A new activates adenylate cyclase of NCB-PO cell membranes to an extent similar to prostacyclin and with a compa- rable high affinity. The binding of [3H]Iloprost to NCB-20 membranes was rapid with an association rate constant (k+d of 2.01 X lo6 M-' s-' at 20 "C. The rate constant for the dissociation of the ligand-receptor complex ( k l ) was 1.19 X lo-' s-', giving a dissociation constant (kl/k+1) of 5.9 nM. The equilibrium dissocia- tion constant was 29.9 nM, and the membranes had a maximum binding capacity of 347 fmol mg-' protein. the molecular of and the in the plasma of the

It has not been possible to solubilize the PGIz receptor without loss of the capacity of the receptor to bind PGIZ. No molecular weight determinations have been made for the receptor, and its structure is unknown. The technique of radiation inactivation has been employed here to determine the molecular weight of the functional PGI, receptor within the plasma membrane of NCB-20 cells. This technique (for reviews see Refs. 15 and 16) has the advantage that membrane-bound receptors and enzymes need not be purified in order to determine their molecular weight. NCB-20 membranes were lyophilized and irradiated in the beam of a linear accelerator in order to examine the components of the PG1,stimulated adenylate cyclase system. Previous reports of radioligand binding to high affinity PGI, receptors have described experiments with [3H]9P-PG12 (17, 18) and [3H] 19), neither of which radioligands are currently available from the usual commercial sources. The synthesis of [3H]11@-PG12 (20) is an exacting task, and the instability of the compound makes it unsuitable in many ways for radioligand binding. A new and stable PGI, analog,

MATERIALS AND METHODS
Cell Culture-Cells of the NCB-20 neuronal hybrid cell line were cultured in Dulbecco's modified Eagle's medium (Gibco Europe) containing either 10% (v/v) newborn calf serum (Gibco Europe) or 5% (v/v) newborn calf serum and 5% (v/v) fetal calf serum (Flow Laboratories). In addition the medium was supplemented with 1 p~ aminopterin (Sigma London Chemical Co. Ltd.), 100 p~ hypoxanthine (Sigma London Chemical Co. Ltd.), and 16 p~ thymidine (Sigma London Chemical Co. Ltd.). The cells were maintained at 37 "C in a humidified atmosphere of 10% COZ in air and were harvested by agitation in Ca2+-and MP-free Dulbecco's phosphatebuffered saline (Gibco Europe). CelI pellets were frozen at -80 "C until required.
Preparation of Cell Membranes-NCB-20 cells were suspended in 25 mM Tris-HC1 buffer, pH 8.5, containing 0.29 M sucrose, and homogenized at 4 "C with a tightly fitting Dounce homogenizer. Homogenate8 containing approximately 7.0 mg/ml of protein were used immediately or frozen at -80 "C until required. To prepare membranes, cell homogenates were centrifuged at 500 X g for 20 min to remove undisrupted cells and nuclei. The membranes remained in suspension and were pelleted by centrifugation at 100,000 X g for 20 min at 4 "C. The membranes were washed 3 times by suspension in 50 mM Tris-HC1 buffer, pH 8.5, and centrifugation was at 100,000 x g. Finally the membranes were resuspended in 25 mM Tris-HC1 buffer, pH 8.5, containing 0.29 M sucrose before lyophilization and irradiation, or for adenylate cyclase experiments. Membranes were resuspended in 50 mM Tris-HC1 buffer, pH 7.4, for experiments to deter-mine [3H]Iloprost binding. All membranes were stored at -80 "C until required.
Reaction mixtures were prepared on ice and then incubated for 20 min at 30 'C. The reaction was terminated by the addition of 4 ml of ice-cold 50 mM Tris-HCl buffer, pH 7.4, with filtration of the membranes through GF/C glass fiber discs (Whatman). The filters were washed three times with the same ice-cold buffer and then dried under an infrared lamp for 1 h. The filters were counted in 10 ml of Instagel (Packard Instrument Co.).
Radiation Znactiuation-NCB-20 membranes were prepared as described and resuspended in the appropriate buffer. Membranes were divided into stoppered polycarbonate tubes in volumes of 450 p1 containing approximately 1.5 mg of protein per tube and frozen at -80 "C prior to lyophilization. Three marker enzymes were used as molecular weight standards. Lysozyme (muramidase, EC 3.2.1.17) from chicken egg white (Sigma London Chemical Co. Ltd.) and N-acety~-~-D-glucosarninidase (EC 3.2.1.30) from jack beans (Sigma London Chemical Co. Ltd.) were either mixed with NCB-20 membranes for irradiation or were irradiated together after suspension in water or buffer followed by lyophilization. Acetylcholinesterase (acetylcholine acetylhydrolase, EC 3.1.1.7) activity was measured in NCB-20 membranes which had been prepared and lyophilized in 25 mM Tris-HC1 buffer, pH 8.5, containing 0.29 M sucrose.
Samples were irradiated using the 7-MeV electron beam produced by the linear accelerator at the MRC Cyclotron Unit, London. The samples were irradiated on ice in a Perspex or polystyrene holder. Doses were given additively in triplicate up to 30 megarads using a dose rate of 400 kilorads/min. The dose given was calculated using a Balowin-Farmer ionization chamber to measure the current. Furthermore, the dose delivered to the sample was checked using lithium fluoride thermoluminescence. After irradiation, samples were stored at -80 "C. Prior to assay triplicate samples were reconstituted in water and surviving enzyme activity or binding capacity monitored in duplicate. Surviving activity was plotted on a log scale against the dose (megarads), and data were regressed (unweighted) as a linear function constrained through the origin ( x = 0, y = 100%) using the Minitab statistics package (24). Molecular weights were determined from the relationship of Kepner and Macey (251, molecular weight = ~ 6.4 X 10" Dm where D37 was the radiation dose (rads) required to decrease the enzyme activity or binding capacity to 37% of its original activity. No temperature correction factor was necessary for the calculation as samples were irradiated at 4 "C rather than at low temperature (26). When a linear relationship was not obtained, the data were fitted to a four-parameter model for biexponential decay using an iterative computer program (27). Lysozyme Actiuity-The method of Shugar (28) was followed. The reconstituted sample was made to 1 ml with distilled water and added to 500 p1 of 0.5 M NaCl and 1.5 ml of Micrococcus lysodeikticus suspension (0.15 mg/ml, Sigma London Chemical Co. Ltd.) in 66.6 mM phosphate buffer, pH 6.24. The decrease in absorbance at 450 nm was measured in a Perkin-Elmer 55s spectrophotometer.
N-Acetyl-8-D-glueosaminidase Actiuity-N-Acetyl-8-D-glucosaminidase activity was measured (29) by the addition of 50 pl of the sample (reconstituted in water) to 450 pl of 4 mM p-nitropheny1-Nacetyl-B-D-glucosaminide (Sigma London Chemical Co. Ltd.) in 40 mM citrate buffer, pH 4.5, containing 5% methanol. Reactions proceeded for 7-10 min at 37 "C and were then terminated by the addition of 1.5 ml of 2 M glycine, pH 10.7. The absorbance at 410 nm was recorded in a Gilford 300N spectrophotometer.
Acetylcholinesterase Actiuity-Acetylcholinesterase activity was measured by a modification of the method of Ellman et aL (30).
Reconstituted membranes (120 pl) were incubated at 37 "C for 3 min in a spectrophotometer cuvette with 780 pl of 100 mM Tris-HC1 buffer, pH 8.0, and 30 pl of 10 mM dithiobisnitrobenzoate (Sigma London Chemical Co. Ltd.) plus 1.79 mM sodium bicarbonate in 100 mM Tris-HC1 buffer, pH 7.0. To this was added 6 pl of 75 mM acetylthiocholine iodide (Sigma London Chemical Co. Ltd.), and the increase in absorbance at 412 nm was measured using a Perkin-Elmer 55s spectrophotometer.
Protein measurements were made using the method of Lowry et al. (31).

RESULTS
Activation of Adenylate Cyclase-A saturating concentration of Iloprost activated adenylate cyclase to an extent similar to PGIz and its stable analog carbacyclin (Fig. 1). Each of the prostaglandins yielded linear Eadie-Hofstee plots (Fig.  1B) suggesting that activation of adenylate cyclase by these prostaglandins is mediated by single receptor populations. The SV,, (maximum enzyme activation) is given by the x intercept and was similar for all 3 prostaglandins (Iloprost = 15.7, PGI, = 18.5, carbacyclin = 15.8 pmol of CAMP min" Binding of [3H]Iloprost to NCB-20 Membranes-The binding at equilibrium of selected concentrations of [3H]Iloprost to NCB-20 membranes is shown in Fig. 2. Specific binding was saturable, and the Scatchard plot (Fig. 2B) showed that B,. (maximum binding capacity of the membranes) was 347 fmol mg" protein, and the equilibrium dissociation constant (Kd) was 29.9 nM.
The rate of association of 15 nM [3H]Iloprost to the PGIz receptor of the NCB-20 membranes was monitored at 20 "C ( Fig. 3). At this concentration of ligand, nonspecific binding was 15% of total binding. A pseudo first-order plot of the data (Fig. 3B)  of cold ligand (Fig. 4). A semi-log plot of the data (Fig. 4B) revealed a half-time (tH) of 584.5 s, from which k-I was calculated to be 1.19 x s-'. The dissociation constant (k-l/k+l) was calculated to be 5.9 nM.
Because of these results, membranes from the NCB-20 cells were suspended in 50 mM Tris-HC1 buffer, pH 8.5, and lyophilized. After reconstitution the membranes retained 88.0% of their basal and PGIz-stimulated adenylate cyclase activity and 100% of their acetylcholinesterase activity. The  (Table 11). The experiments yielded high molecular weights which were not in agreement with previous reports. Under these conditions, two target sizes were obtained for C of 166,000 and 842,000; CN yielded a molecular weight of 430,000 and CNR was 904,000. Similar results were obtained when membranes were prepared and lyophilized in a high salt buffer (50 mM Tris-HC1 buffer, pH 8.5, containing 100 mM NaC1). Experiments were designed to investigate further the effects of salt concentration and solute molarity on the target sizes of C, CN, and CNR. When membranes were suspended in 25 mM Tris-HC1 buffer, pH 8.5, containing 0.29 M sucrose and lyophilized, irradiation yielded much smaller molecular weights (Table 11), all of which were single target sizes derived from simple exponential decay curves. The molecular weight of acetylcholinesterase found by irradiation of NCB-20 membranes under these conditions was 62,000 ( r = 0.997).
The binding was measured of [3H]Iloprost to irradiated NCB-20 membranes prepared and lyophilized in the Tris-HC1 buffer containing sucrose. In three separate experiments simple exponential decay curves were obtained, yielding a molecular weight value (mean f S.E.) of 82,800 f 12,900, The results of one of these experiments are shown in Fig. 5.

DISCUSSION
Direct evidence of a protein receptor for PGIp on NCB-20 cells has been presented. The receptor is sensitive to trypsin, and incubation of the membranes with 0.5 mg ml-' trypsin for 20 min at 30 "C reduces specific binding of [3H]Iloprost to 24% of control values (data not presented). Iloprost is a novel ligand which activates adenylate cyclase to an extent similar to the natural ligand PGI, and another synthetic PGI, agonist, carbacyclin.
Iloprost has previously been shown to mimic the biological activity of PGIz in its action on platelets and bovine coronary arteries (33)(34)(35).
[3H]Iloprost binds to specific PGI, receptors on platelets (36), bovine coronary artery (37), and porcine aorta (38). The report of binding to porcine aorta differed from this and other studies using [3H]Iloprost or [3H]PGI, (11,(17)(18)(19) in that the data presented were interpreted as indicative of positive cooperativity. [3H]Iloprost has been employed in these studies and is well suited as a radioligand because of its relative stability compared to PGI, and its low nonspecific binding. The maximum binding capacity of [3H] Iloprost in NCB-80 membranes is similar to the maximum [3H]PGIz binding ( l l ) , and the rank order of potency of selected prostaglandins in displacing bound [3H]Ilopr~~t (Table I) and [3H]PGIz is identical (11). These results correlate well with the order of potency of each prostaglandin for adenylate cyclase activation (12).
To date, no reports have been published of successful solubilization of the PGI, receptor with retained capacity of the receptor to bind PGI,. However, previous studies have suggested the presence of high-molecular-weight receptors for other prostaglandins. Rat liver plasma membranes have been prelabeled with [3H]PGEI and solubilized in Triton X-100, indicating a molecular weight of 105,000 by gel filtration of the putative ligand-receptor complex (39). Similarly the PGF2, receptor of bovine corpora lutea has been prelabeled and solubilized (40) indicating a molecular weight of 107,000. In these studies, however, it was not possible to verify the presence of the solubilized receptor with certainty, as the capacity for ligand binding was lost. In addition, prostaglandins generally have high values for nonspecific binding, and hence prelabeling of the membranes may yield radioligand coupled to proteins other than the "specific" membrane receptors.
The physiochemical characteristics of many receptors have been examined after solubilization, for example the opiate (41), insulin (42), and al-and /3-adrenergic (43,44) receptors. Similar results have not been achieved with any prostaglandin receptor, which prompted the present study to examine the PGI, receptor in its native membrane environment by the technique of radiation inactivation. Preliminary results with marker enzymes indicated a need for buffering during lyophilization to avoid what seemed to be aggregation of the enzymes. The molecular weights obtained for the enzymes lyophilized in water were greater than the values reported previously, based on hydrodynamic and radiation methods (45)(46)(47). For this reason NCB-PO membranes were prepared for lyophilization and irradiation by suspension in Tris-HC1 buffer. It has been suggested (48), in the case of acetylcholinesterase, that Tris may minimize the spread of electrons from one polypeptide chain to another during irradiation. Under these circumstances the target size obtained would be that of the smallest functional unit. However, the irradiation of NCB-20 membranes in Tris-HC1 buffer reported here yielded molecular weights of C and CN much greater than those determined from hydrodynamic studies (Table 11).
Measurements by conventional methods of the GTP-sensitive catalytic subunit (CN) have yielded molecular weights between 160,000 and 220,000 from rat renal medulla (49), canine cerebral cortex (50), mature rat testis (51), and S49 lymphoma cells (52). In some of these reports smaller molecular weight subunits have also been described. A component of the regulatory subunit (N) was first identified as a 42,000 polypeptide labeled by GTP and cholera toxin (53). Subsequent measurement of the molecular weight of the complete N subunit yielded a value of approximately 130,000 in erythrocyte membranes (54) and S49 lymphoma cells (55). In radiation inactivation studies, C has been assigned a molecular weight of 150,000 in frozen and lyophilized hepatic membranes (56) and 92,000 in frozen turkey erythrocytes (32). The CN complex has yielded molecular weights of 230,000 in lyophilized or frozen hepatic membranes (56), 226,000 in frozen turkey erythrocytes (32), and 160,000 in lyophilized hepatic membranes (57).
Basal enzyme activity after irradiation was assayed in NCB-20 membranes using MnATP as substrate. It has been shown previously that the uncoupled catalytic subunit uses this substrate preferentially (58,59). In these experiments a nonlinear decay curve was obtained in the measurement of the molecular weight of C from membranes prepared and lyophilized in Tris-HC1 buffer. The high molecular weights of C and the nonlinear decay curve are most readily explained as aggregation of the catalytic subunits during the experimental procedure. Finally, the molecular weights of acetylcholinesterase and the C, CN, and CNR complexes were determined in membranes suspended and lyophilized in Tris-HC1 buffer containing sucrose. The molecular weight obtained for acetylcholinesterase was 62,000. Acetylcholinesterase has been shown to exist in several different forms (60), but in the central nervous system of mammals and birds only globular forms are present. They have a subunit molecular weight of 71,000 which may form dimers and tetramers. Radiation inactivation of membrane-bound acetylcholinesterase from erythrocyte ghosts has yielded a molecular weight of 75,000 (61) from frozen or lyophilized preparations in one study and 56,000 for lyophilized preparations in another.
The molecular weight values obtained for the components of adenylate cyclase were smaller (C = 111,000, CN = 200,000, CNR = 283,000) when the membranes were prepared in Tris-HC1 buffer containing sucrose and corresponded well to the published values for C and CN. Under these experimental conditions single target sizes were obtained in all cases. Sucrose has been reported previously to increase the radiation sensitivity of enzymes in the dry state (62). The molecular weight of the PGI, receptor was also determined by measurement of the decline of [3H]Iloprost binding with increasing radiation doses (Fig. 5). Membranes were again prepared and lyophilized in Tris-HC1 buffer containing sucrose. The target size (mean * S.E.) was calculated to be 82,800 k 12,900 ( n = 3). This was in good agreement with the molecular weight of 83,000 derived from the adenylate cyclase data (calculated as the difference between the molecular weights of CNR and CN).
All experiments in this study were performed on lyophilized samples which avoided the necessity of a temperature correction factor. In the presence of sucrose, the molecular weight values obtained for acetylcholinesterase, C, and N corresponded closely to results obtained by irradiation of frozen membranes or by hydrodynamic methods.
In conclusion, the molecular weight of the PGI, receptor obtained by monitoring the loss of the [3H]Iloprost binding capacity correlated well with the result derived from the adenylate cyclase data. Furthermore, the results obtained for the molecular weights of acetylcholinesterase (a membranebound enzyme) and for the regulatory and catalytic subunits of adenylate cyclase were similar to those reported in the literature. These points all serve to increase the confidence with which the molecular weight of the PGI, receptor is presented. However, it should be noted that conditions that yielded the lowest molecular weights were selected as the preferred experimental method, and the remote possibility still exists that the high-molecular-weight aggregates may be the functional or active structures within the membrane.