key: cord-0005053-kggbh8c7 authors: Uhlén, Staffan; Xial, Yun; Chhajlanil, Vijay; Lien, Eric J.; Wikberg, Jarl E. S. title: Evidence for the existence of two forms of α(2A)-adrenoceptors in the rat date: 1993 journal: Naunyn Schmiedebergs Arch Pharmacol DOI: 10.1007/bf00167446 sha: 01c9dcd5cdb878c778dc7dbced9945c3b9ca8a42 doc_id: 5053 cord_uid: kggbh8c7 The α(2A)-adrenoceptors in rat spleen, kidney, spinal cord and cerebral cortex were studied using [(3)H]-RX821002 radioligand binding. In the spleen, spinal cord and cerebral cortex, the ligand bound to saturable sites with a K (d) of about 1 nmol/l and capacities of 134, 240 and 290 fmol/mg protein, respectively. Computer modelling competition curves for 39 drugs, including those for α(2A)-, α(2B)- or α(2C)-adrenoceptor selective drugs, indicated that the sites labelled by [(3)H]-RX821002 in the spleen consisted of a single population of α(2A)-adrenoceptors. However, the competition curves for guanoxabenz were definitely biphasic and resolved into two site fits, indicating that guanoxabenz was binding to both high affinity (K (d) = 35 nmol/1) and low affinity (K (d) = 8900 nmol/1) α(2A)-adrenoceptor sites in the proportions 57% and 43%, respectively. The K (d) (S)for a number of α(2)-adrenoceptor subtype selective drugs, measured in competition with [(3)H]-RX821002 in cerebral cortex and spinal cord, were highly correlated with those obtained in the spleen indicating their α(2A)-adrenoceptor nature. However, by contrast to the results with the spleen, the guanoxabenz competition curves for the spinal cord and cerebral cortex were monophasic and resolved only into one site fits, the K (d) of guanoxabenz being about 4000 nmol/l for both tissues. Drug K (d) (S)for kidney α(2A)-adrenoceptors were also determined using [(3)H]-RX821002. For nearly all drugs tested, the K (d) (S)were highly correlated with those found for the α(2A)-adrenoceptors in the other rat tissues. However, for guanoxabenz, the data indicated that it competed with [(3)H]-RX821002 at a single α(2A)-adrenoceptor site with a K (d) of 39 nmol/1. When the rat α(2A)-adrenoceptor gene RG20 was transiently expressed in COS-7 cells and its ligand binding properties probed using [(3)H]-RX821002, the drug K (d) (S)obtained were also highly correlated with those found for the α(2A)-adrenoceptors in the spleen, cerebral cortex, spinal cord and kidney of the rat. For the RG20 encoded receptor, the guanoxabenz competition curves were steep and monophasic and modelled best into one site fits, with the Kd of guanoxabenz being 5200 nmol/1. It is suggested that guanoxabenz can differentiate between two forms of α(2A)-adrenoceptors in the rat: α(2A1) and α(2A2). The α(2A1)-form is present in the spleen and kidney where it shows a high apparent affinity for guanoxabenz. The α(2A2)-form shows a low apparent affinity for guanoxabenz and is present in the spleen, cerebal cortex and spinal cord. The α(2A2)-form of the rat α(2)-adrenoceptor appears to be encoded by the RG20 gene. The α(2A), and α(2A2)-adrenoceptor forms do not represent high and low affinity receptor forms for agonists because assays included EDTA, Gpp(NH)p and Na(+), which eliminated the high affinity receptors for agonists. In some recent studies using radioligand binding we showed that, in the rat, there are at least three distinct a2-adrenoceptor subtypes present (Uhl6n and Wikberg 1991a-c; Uhl6n et al. 1992; Xia et al. 1993) . Following the earlier proposition of Bylund (1988 Bylund ( , 1992 for the nomenclature of aE-adrenoceptors, these rat receptor were classified as being a2A, a2B and a2c-adrenoceptors. The aEA-adrenoceptors were found in the kidney, spinal cord and cerebral cortex, the a2B-adrenoceptors in the kidney and neonatal lung and the a2c-adrenoceptors in the spinal cord and cerebral cortex. However, we also found strong evidence that the rat a2a-adrenoceptors were heterogenous and that they could seemingly be subdivided into two forms which we termed a2B~ and a2B2 Xia et al. 1993 ). Among several subtype-selective compounds, guanoxabenz was shown to be the best drug to differentiate between the aEm-and aEBE-adrenoceptors. In a previous study, we also noted that guanoxabenz showed grossly different affinities for the a2Aadrenoceptors in the kidney a n d in the cerebral cortex of the rat, possibly indicating some type of heterogeneity a m o n g a2A-adrenoceptors (Uhl6n et al. 1992) . Th.e present study was designed to further characterize a2nadrenoceptors in the rat. We now present data indicating that rat a2A-adrenoceptors also seem to exist in two forms: a2A 1 a n d aZA2. The a2A 1 form was f o u n d in the spleen a n d kidney whereas the a2A2 form was found in the spleen, spinal cord and cerebral cortex. Moreover, when the recently cloned rat a2A-adrenoceptor gene R G 2 0 (Lanier et al. 1991 ) was used to express a2-adrenoceptors in COS-7 cells, the receptors obtained showed the properties of the a2A2-form of the a2Aadrenoceptor. So far, the only drug which is capable of distinguishing between the a2A 1-a n d a2A2-adrenoceptor forms appears to be guanoxabenz, its affinity differing by approximately 100-fold for the two forms. Expression of RG20 in COS-7 cells. The RG20 gene was cloned into the EcoR 1 -Not 1 site of the pMT3 vector as previously described (Lanier et al. 1991 ). The plasmid was purified using the Quiagen kit before using it for transfeetion into COS-7 cells to afford its transient expression. COS-7 cells were grown in Dulbecco's modified Eagle medium with 10% fetal calf serum. Subconfluent cultures in 60 mm dishes were transfected with 1 ~tg of plasmid DNA and 30-50 ~tg Lipofectin reagent (BRL, USA) according to the instructions supplied by the manufacturer. Cells were harvested 48-60 h after the transfection for preparation of membranes. Membrane preparations. Membranes from rat spleen, kidney, spinal cord and cerebral cortex were prepared from Sprague-Dawley rats, essentially as described previously (Uhl6n and Wikberg 1991 b) . The final membrane fractions were diluted, to give protein concentrations of 2.4 mg protein/ml for kidney and -1.2 mg protein/ml for spleen spinal cord, and cerebral cortex, with 1.5 mmol/1 EDTA-50 mmol/1 Tris-HC1 (pH 7.5). The COS-7 cell membranes were prepared by scraping cells into ice-cold phosphate buffered saline containing 0.54 mmol/1 EDTA, pH 7.2. After centrifugation at 800×g for 10 rain, the cells were resuspended in ice-cold 50 mmol/1 Tris-HC1 containing 5 mmol/1 ED-TA, 0.1 mmol/1 phenylmethylsulphonyl fluoride, 10~tg/ml soybean trypsin inhibitor and 200 ~tg/ml bacitracin, pH 7.5, and homogenized 3 × for 15 sec with an Ultra-Turrax T25 at 24000 rpm. The homogenates were then spun at 38000 g for 20 rain and the final pellet resuspended in 1.5 retool/1 EDTA, -50 mmol/1 Tris-HC1, pH 7.5 to give a protein concentration of -1.2 mg/ml. Membrane preparations were frozen and stored at -80°C for up to 14 days before use. Protein was measured according to Lowry et al. (1951) . Binding studies. Radioligand binding assays were done, essentially as previously described (Uhl6n and Wikberg 1991a) , by incubating 120-240~tg of the membranes in 1501xl of a solution containing 1 retool/1 EDTA, 100 ~tmol/1 Gpp(NH)p (guanyl-5'-yl-amido-diphosphate), 140mmol/1NaC1, 33mmol/1 Tris-HC1, pH7.5 with [3H]-RX821002 and different drugs for 1 h at 25 °C and then filtering and washing on Whatman GF/C filters. All assays were performed in duplicate and saturation experiments included 12 concentrations of [3H]-RX821002. Non-specific binding was determined in the presence of 1 ~tmol/1 BDF 8 933. Computer modelling of the data was as previously described (Uhl6n and Wikberg 1991 a, c) , which gave the dissociation constants (Kds) of drugs. The PKi-values for the drugs were then calculated as the -logl0(Kd). Hill coefficients were calculated by fitting the data to the four parameter logistic function using non-linear regression. Experimentally determined values are given as the mean_+SEM. Isotopes, drugs and chemicals. [3H]-RX 821002 (1,4-(6,7(n) -3H)benzodioxan-2-methoxy-2-yl)-2-imidazoline, (51 Ci/mmol) was from Amersham; (-)-adrenaline, amiloride, (-)-noradrenaline, dopamine, chlorpromazine, corynanthine, prazosin and yohimbine were from Sigma Chemical Co.; (+)-adrenaline was from Sterling-Winthrop Research Institute, Rensselaer, NY; ARC239 (2-(2,4-(O-methoxyphenyl)-piperazin-1-yl)-ethyl-4,4-dimethyl-1,3 (2 H, 4 H)-isoquinolindione) and azepexole (formerly known as BHT 933) from Thomae, Biberach, Germany; benoxathian and WB 4101 from Research Biochemicals, Natick, Mass.; BDF 8933 (4-fluoro-2-(imidazoline-2-ylamino)-isoindoline maleate) from Beiersdorf, Hamburg, Germany; BRL44408 (2-[2H-(1-methyl-l,3-dihydroisoindole)methyl]-4,5-dihydroimidazole) and BRL 41992 ( 1,2-dimethyl-2,3,9,13 b-tetrahydro-1H-dibenzo [c, f] imidazol[1,5-a]azepine) were from Beecham, Essex, UK; clonidine from Boehringer Ingelheim, Ingelheim/Rhein, Germany; FLA 151 (2,6-dichlorobenzylidene-amino-3,3-dimethylguanidine) and FLA 163 (2-chlorobenzylideneamino-3,3-dimethylguanidine) were a kind gift from Dr. Lennart Florval, Astra, SOdert/~lje, Sweden; guanfacine and guanabenz were gifts from Dr. Claes Post, Astra, S6dertfilje, Sweden; guanoxabenz and RU24969 (5-methoxy-3-(1,2,3,6,-tetrahydropyridin-4-yl)-lH-indole) were from Roussel, Romainville, France; methysergide from Sandoz, Basel, Switzerland; oxymetazoline from Draeo, Lund, Sweden; ICI 106,270 (1,6-(2-ehloro-6-fluorophenyl)-2,3,6,7-tetrahydro-5H-pyrrolo-[1,2-a]-imidazole) was from Imperial Chemical Industries PLC, Macclesfield, Cheshire, UK; rauwolscine from Roth, Karlsruhe, Germany, rilmenidine from Servier, Neuilly-sur-Seine, France; (+) and (-)mianserine from Organon, Oss. Holland, MK-912 (L-657,743) from MSD, Rahway, NJ; SKF 104078 (6-chloro-9-[3-methyl-2-butenyloxy]-3methyl-lH-2,3,4,5-tetrahydro-3-benzazepine) from SK&F, Swedeland, PA; UK 14,304 (5-bromo-6-(2-imidazoline-2-ylamino)-quinoxaline) from Pfizer, Sandwich, UK; Wy 26,392 (N-((2fl, 11ba)-l,3,4,6,7,11bhexahydro-2H-benzo-(a)-quinolizin-2-y)l-N-methylpropanesulphonamide) from Wyeth, Maidenhead, Berks., UK; LW01 (1-(9'-chloroanthrylmethylene)amino-3-hydroxy-guanidine tosylate), LW03 (1-(2'chloro-4',5'-methylenedioxybenzylidene) amino-3-hydroxyguanidine tosylate), LW04 (l-(Y,4'-ethylene-dioxybenzylidene)amino-3-hydroxyguanidine tosylate), LWll (1-(2',hydroxybenzylidene)amino-3-hydroxyguanidine tosylate), LW 12 (I-(3'-hydroxypyridinyl-methyleno)amino-3hydroxyguanidine tosylate), LT07 (1-[[3-(hexyloxy)benzylidene]amino]-3-hydroxyguanidine tosylate), LT 11 (1-[(3-methoxy-benzylidene)amino]-3-hydroxyguanidine tosylate), ATL26 (1-[[4-(trifluoromethyl)benzylidene]amino]-3-hydroxyguanidine tosylate) were synthesized by Drs. P.-H. Wang, A.W. Tai and A. Tang in the laboratory of one of the authors (E.J.L.), as described (Tai et al. 1984; Tang et al. 1985; Wang et al. 1990 ). Quiagen kit was from Quiagen, USA, Lipofectin reagent, Dulbecco's modified Eagle medium and fetal calf serum was from BRL, USA. All other chemicals were purchased from Merck or Sigma and were of analytical quality. The b i n d i n g of [3H]-RX 821002 to rat spleen cell membranes was characterized by incubating different concentrations of the radioligand in the absence and in the presence of 1 ~tmol/1 of BDF 8933, the latter being used to define non-specific binding. The resulting curves ( Table 1 , which shows the pKi-values obtained for the drugs as well as the Hill coefficients (~H) of the competition curves. All drugs, except guanoxabenz, yielded competition curves which were steep and monophasic and which showed Hill coefficients close to unity. The computer modelling for all these 39 compounds showed that the data fitted a one-site model best. Fitting the data to a two-site model resulted in only marginal and statistically insignificant (P>0.05) reductions in the sums of squares when compared with the sums of squares for a one-site model. Notably, strong agonists such as (-)-adrenaline and (-)-noradrenaline also gave steep curves which were resolved only into one-site fits. This indicates that the assays, which included EDTA, Gpp(NH)p and Na +, had completely eliminated the agonist high affinity az-adrenoceptor form. The pKi-values obtained for the a2h-adrenoceptor selective drugs oxymetazoline, BRL 44408 and guanfacine, the a2B-adrenoceptor selective drugs prazosin and ARC 239, as well as the azcadrenoceptor selective compounds MK912, WB4101 and rauwolscine (see Uhl6n and Wikberg 1991 c; Uhl6n et al. 1992) were fully compatable with the notion that the sites labelled by [3H]-RX 821002 in the rat spleen were aZAadrenoceptors. However, the competition curves for guanoxabenz were strongly biphasic ( Fig. 2A ) and computer modelled best into two-site fits, the analysis showing that two site fits resulted in drastic and highly significant (P<0.0001) reductions in the sums of squares as compared to the values obtained for one site fits for all tests, whereas three sites fits did not improve the regressions. The analysis thus indicated that guanoxabenz was bound to a high affinity site (Kd = 35.5 nmol/1) and also to a low affinity site (Kd = 8910 nmol/l), a difference in affinities amounting to about 250-fold. The analysis further showed that the proportion of high affinity sites for guanoxabenz was 57.2070 +2.4070 and that of the low affinity sites was 42.8070+2.4070 (n =26). The reason that guanoxabenz was tested 26 times was that a competition curve for guanoxabenz was included daily in all assays when the other 39 compounds were tested. This was done in order to ascertain that the two forms of a2Aadrenoceptors were not missing from some of the batches of membranes used and that day to day variations did not result in an inability to observe putative differences in af- Competitor Log(M) (11) with the A spleen, B kidney, C spinal cord, and D cerebral cortex of the rat. The competition curves for binding in the kidney were obtained by using -2 nmol/1 [3H]-RX 821002 and a fixed concentration of 1 ~mol/1 ARC 239 for all assays. For spinal cord and cortex, a fixed concentration of 1.7 nmol/1 of [3H]-RX821002 was used. The lines represent the computer-drawn best fits assuming that the ligands bound to independent sites according to the law of mass action. For the spinal cord and cerebral cortex, the data were fitted to a model that assumed one site to be present. For the spleen and kidney, the model used assumed two sites to be present. In all panels the ordinates represent the total binding aK d of guanoxabenz for high affinity site in spleen b Kd of guanoxabenz for low affinitiy site in spleen finities for any of the other 39 drugs tested for binding to the two forms of the a2A-adrenoceptor in the spleen. The data showed that the strongly biphasic competition curve for guanoxabenz was consistently present in all 26 tests. As summarized in Table 1 , a number of structural analogs of guanoxabenz were included among the 40 compounds tested on the spleen. Since all of these gave monophasic competition curves which modelled best into one site fits, the ability of guanoxabenz to distinguish between the two forms of the spleen a2A-adrenoceptor appears to be unique among the 40 compounds tested. In one of our previous studies (Uhl6n and Wikberg 1991 a), we used an elaborate 6-curve assay which was designed to obtain simultaneously binding constants of drugs for the azA-, aZBl-and a2Bz-adrenoceptors that are present in the rat kidney. In the present study, we were only interested in characterizing a2A-types of a2-adrenoceptors. We therefore developed a simplified assay to obtain drug pKi-values for the kidney aaA-adrenoceptor. Our previous study showed that ARC 239 had a high affinity for the kidney a2B 1-and a2R2-adrenoceptors but a low affinity for the azA-adrenoceptor. Theoretical calculations, using the Kd-Values for [3H]-RX 821002 and ARC 239 given in our previous paper (Uhl6n and Wikberg 1991 a) indicated that if 1 gmol/1 of ARC 239 was included in the assay, the binding of 2 nmol/1 [3H]-RX821002 to the a2B 1-and a2B2-receptors would be blocked by 99O7o and 95O70, respectively, whereas only 8°7o of the binding to the a2A-receptors would be blocked. To evaluate this approach, we obtained competition curves for prazosin, guanoxabenz and oxymetazoline, which are subtype-selective drugs for the three kidney a2-adrenoceptors (Uhl6n and Wikberg 1991a), using 2nmol/1 [3H]-RX 821002 as well as 1 gmol/1 ARC 239 in the assays. In addition, a full competition curve for ARC 239 was obtained using the same conditions (Fig. 2 B) . When the data were analyzed by computer modelling, it was found that the drugs tested gave significant two site fits (P<0.001). These results indicated that, despite the masking effect of 1 ~mol/1 ARC239, some of the a2B2-adrenoceptors interfered in the assay. The interference from these receptors can be clearly seen as the minor tail on the competition curves of oxymetazoline and guanoxabenz as well as the minor distortion of the low concentration range of the prazosin and ARC 239 competition curve in Fig. 2B . The calculations showed that the sites labelled by [3H]-RX 821002 corresponded to -91% of a2A-sites and -9% of a2B2-sites , which was in full accord with the theoretical calculations. When the pKi-values of drugs were calculated using this form of the assay, the data were, therefore, computer modelled into two site fits to ensure that the azn2-adrenoceptor did not interfere in the determination of the drug PKi-values for the kidney azA-adrenoceptor (see legend to Table 2 for details of these calculations). Using this approach, we determined the binding constants for some compounds in addition to prazosin, guanoxabenz, oxymetazoline and ARC 239. The data from all these calculations are given in Table 2 along with the drug pKi-values determined for the rat kidney a2A-adrenoceptor in our previous study using the more elaborate 6-curve assay (Uhl6n and Wikberg 1991a) . The PKi-values for guanoxabenz, oxymetazoline, prazosin and ARC 239 obtained using the new approach are practically the same as the pKi-values obtained in our previous study, indicating the validity of the new method. As can be seen from Table 2 , the drug pKi-values obtained for the kidney a2A-adrenoceptors are virtually the same as those obtained with the rat spleen indicating that the receptors in both tissues belong to the a2A-adrenoceptor category. However, in contrast with the results obtained with the spleen, the analysis of the data for guanoxabenz binding to kidney membranes indicated that guanoxabenz bound to a single aZAadrenoceptor site; the pKi-value being practically the same as that obtained for the high affinity site in the spleen (Table 1 , 2; c.f. Fig. 2A, B) . The other N-hydroxyguanidine analogs bound with practically the same affinities to the kidney and spleen azA-adrenoceptors, which further demonstrates that guanoxabenz is unique, among the substances tested, in its ability to differentiate two forms of the azA-adrenoceptor in the rat. Table 2 . As can be seen from the table, all drugs except guanoxabenz gave pKi-values which are similar to those obtained with the spleen and with the kidney supporting the notion that all the receptors studied were of the azA-type. However, the affinity of guanoxabenz for the spinal cord azA-adrenoceptors was 100-fold lower than that obtained for the receptors in the kidney. Moreover, the affinity in the cord was also about 100-fold lower than the affinity obtained for the high affinity azA-adrenoceptor site in the spleen. On the other hand, the PKi-value for guanoxabenz interacting with the spinal cord aza-adrenoceptor was close to that obtained for the low affinity site in the spleen. These differences in drug affinities are shown in Fig. 2 . Competition curves for oxymetazoline, guanoxabenz, prazosin and ARC 239 obtained with the spinal cord are shown in Fig. 2C . Competition curves for the same compounds obtained with kidney are shown in Fig. 2B . It can be seen that, with the spinal cord, the competition curve for guanoxabenz is located far to the right of the other curves, indicating that guanoxabenz has the lowest affinity of the four drugs tested for the spinal cord a2Aadrenoceptor. With the kidney, however, the competition curve for guanoxabenz is located far to the left of the prazosin and ARC 239 competition curves and is aligned just to the right of the oxymetazoline curve, which indicates that guanoxabenz is almost as potent as oxymetazoline in binding to the kidney azA-adrenoceptors. As can also be seen from Fig. 2A , the low affinity component of the guanoxabenz curve for the spleen in located to the right of the ARC 239 curve. A similar result for the guanoxabenz curve was obtained with the spinal cord (Fig. 2C) . The high affinity component of the guanoxabenz curve for the spleen, on the other hand, is located just to the right of the oxymetazoline curve. This location of the guanoxabenz curve is the same as is found with the kidney (c. f., Fig. 2A and B) . In our previous study (Uhl6n and Wikberg 1991b) we found that [3H]-RX821002 labelled a homogeneous population of azA-adrenoceptors in the rat cerebral cortex. To obtain data for comparison we also evaluated some selected drugs with the cerebral cortex. Preliminary analysis of saturation curves of [3H]-RX 821002 indicated that the ligand labelled a single populations of sites with a Ka of 1.08_+0.02nmol/1 and a Bma. of 289+ 14 fmol/mg protein (n = 2) (data not shown graphically). Competition curves for oxymetazoline, guanoxabenz, prazosin and ARC 239 were obtained as shown in Fig. 2D . The pattern for the competition curves obtained in the cortex is identical to that for the spinal cord. Computer modelling clearly indicated that one-site fits were the most appropriate to describe the experimental data. The pKi-values obtained from the calculations are shown in Table 2 . As can be seen from the table, the drug pK ivalues obtained with the cortex were close to those found with the spinal cord. As with the spinal cord, guanoxabenz showed about 100-fold lower affinity for the cortex azA-receptors than for the kidney azA-adrenoceptors or the high affinity a2A-adrenoceptors in the spleen. Computer modelling of saturation curves, obtained by using [3H]-RX 821002 with membranes prepared from COS-7 cells transiently expressing the RG20-adrenoceptot, showed that the ligand bound a single saturable site with a K d of 0.82+0.03nmol/1 and capacity of 1600 + 63 fmol/mg protein (n = 4) (Fig. 3 B) . Moreover, the Scatchard transforms of the saturation curves were straight, a result which supports the notion that [3H]-RX821002 labels a single site (Fig. 3 A) . Control experiments with membranes from COS-7 cells which had not been treated with the RG20 showed an almost negligible non-specific binding for [3H]-RX821002 (data not shown). Competition curves for drugs obtained using 1.6 nmol/1 [3HI-RX 821 002 were monophasic and best modelled into one site fits (Fig. 3C) . The competition 0.14- , prazosin (©) and guanoxabenz (1) obtained using a fixed concentration. (-1.6 nmol/1) of [3H]-RX821002. The lines represent the computer-drawn best fits assuming that the ligands bound to a single site. The ordinate in C represents the total binding curves for guanoxabenz were also monophasic and also modelled best into one site fits (Figs. 3 C) . A comparison of Fig. 3 C with Fig. 2 reveals that the pattern obtained is similar to that found in spinal cord and cerebral cortex since the guanoxabenz competition curve islocated to the right of the prazosin and ARC 239 competition curves. The PKi-values obtained for a number of drugs with the RG20 encoded a2-adrenoceptor are given in Table 2 . The values are close to those found for the a2A-adrenoceptots in the different rat tissues investigated. The affinity of guanoxabenz for the RG 20-adrenoceptor is similar to the affinity of guanoxabenz for the a2a-adrenoceptors in the spinal cord and cerebral cortex as well as for the low affinity aza-adrenoceptor in the spleen. In the present study we have shown that the rat kidney a2A-adrenoceptors show grossly different binding prop-erties for guanoxabenz when compared with the a2Aadrenoceptors in the spinal cord and cerebral cortex. Thus, guanoxabenz has a Kd-value of --40 nmol/1 for the kidney a2A-adrenoceptor whereas it has Kd-values of about 4,000 nmol/1 for the spinal cord and cerebral cortex a2A-adrenoceptors, respectively. Moreover, the data of the present study show that guanoxabenz apparently binds with two affinities to a2A-adrenoceptors in the rat spleen. The K d of guanoxabenz for the high affinity form of the spleen a2A-adrenoceptor was 35 nmol/1 whereas the K d for the low affinity form was 8,900 nmol/1. These values amount to an approximately 100-to 250-fold difference in apparent affinity for the two forms of the a2A-adrenoceptor. Thus, it appears that one of the two forms of the spleen a2A-adrenoceptor shows a high affinity for guanoxabenz similar to that shown by the a2A-adrenoceptor in the kidney whereas the other shows a low affinity for guanoxabenz similar to that shown by the a2A-adrenoceptors in the spinal cord and cerebral cortex. In the present study we have also shown that the rat a2A-adrenoceptor clone RG20, when expressed in COS-7 cells, produces an a2A-adrenoceptor whose guanoxabenz affinity corresponds exactly to that found for the form of the a2A-adrenoceptor showing the lower affinity for guanoxabenz in the different tissues investigated. The affinities for a number of other drugs are completely consistent with the view that all of the a2-adrenoceptors studied were of the a2A-type. Thus, the affinities for a2A-adrenoceptor selective drugs such as guanfacine, BRL44408 and oxymetazoline, for a2Badrenoceptor selective drugs such as ARC239 and prazosin as well as for a2c-adrenoceptor-selective drugs such as MK 912, WB 4101 and rauwolscine corresponded exactly to those for a2A-adrenoceptors (see Uhl6n and Wikberg 1991 a, c; Uhl6n et al. 1992) . By contrast, these affinities are distinctly different from the affinities that we have previously determined for a2B-and a2cadrenoceptors (Uhl6n and Wikberg 1991 a, c; Uhl~n et al. 1992; Xia et al. 1993) . Therefore, all the az-adrenoceptors investigated in the present study are clearly of the a2A-adrenoceptor type. However, the data of our present study indicate that these a2A-adrenoceptors exist in two forms with highly differing apparent affinities for guanoxabenz. We suggest that these forms of a2Aadrenoceptors should be operationally termed a2A 1 and a2A2, the a2A~ form showing high affinity, and the a2A2 form showing low affinity, for guanoxabenz. This nomenclature is in line with our previous nomenclature for two apparent forms of a2B-adrenoceptors in the rat which were termed a2B1 and a2B2 . Guanoxabenz, which is an N-hydroxyguanidine, seems to have quite remarkable properties that enable it to differentiate between the a2A1-and a2A2-forms of the a2A-adrenoceptors (present study) as well as between the a2B 1-and a2B2-forms of the a2B-adrenoceptors Xia et al. 1993 ). In the present study we evaluated several other N-hydroxyguanidines which are structural analogs to guanoxabenz. Interestingly none of these were capable of delineating the rat q2A1-and aZA2-adrenoceptor subtypes. These results indicate that the structural requirements for selectivity at a2A 1-and a2A2-adrenoceptors are strict and not solely dependent on the hydroxyguanidinium side chain present in guanoxabenz. In this context, we would like to mention that we have recently found that LTI1, which is also an Nhydroxyguanidine, was useful in discriminating between a2B 1-and azB2-adrenoceptors in the rat kidney (Xia et al. 1993 ). In the present study, however, LT 11 failed to distinguish a2A I-from a2A2-adrenoceptors (Table 1 and 2). The molecular basis for the apparent heterogeneity among rat a2A-adrenoceptors is at present not clear. The most straightforward explanation is that azA-adrenoce ptors exist as two distinct species that are possibly coded for by different genes. This interpretation is supported by 'the observation that RG20 encodes a receptor which shows properties similar to those of the a2A2-adrenoceptor of rat tissues. However, other possibilities should also be considered. The a2A-adrenoceptors could, for example, be post-translationally modified so that some of them lost or aquired the ability to bind guanoxabenz with high affinity. Such a modification could involve a specific amino acid which interacts with guanoxabenz in the ligand binding pocket of the a2A-adrenoceptor. A prerequisite for this hypothesis is that the general structure of the receptor is not changed by this modification because none of the other 39 drugs tested were capable of the a2A1/a2Az-delineation. It should be pointed out that these possibilities are highly hypothetical and would have to be supported by the finding of conditions which give the expressed RG20 an affinity for guanoxabenz which corresponds exactly to that for the a2Al-adrenoceptor. On the other hand, the molecular cloning of a distinct and novel rat a2-adrenoceptor gene which, when expressed, yields a receptor with the a2Al-adrenoceptor properties is required to prove the hypothesis that a2A 1and a2A2-adrenoceptors are coded for by two different genes. Besides the RG20 gene in the rat, a number of other a2-adrenoceptor genes have been cloned. Of these the RGI0/pA2d genes (Lanier et al. 1991; Voigt et al. 1991) clearly code for an a2c-type of adrenoceptor (Uhl6n et al. 1992) . The RNG gene appears to code for an a2Btype of adrenoceptor (Zeng et al. 1990 ). Thus none of these rat genes are candidates for a putative a2Al-adrenoceptor gene. Chalberg and coworkers (1990) have cloned a gene, cA2-47, which is almost identical with RG20 albeit with minor sequence differences. The meaning of these minor differences are at present not clear but should prompt further investigations. At present only limited data are available regarding the pharmacological properties of the cA2-47 encoded receptor. The consistency of the method used in the present study to determine drug affinities for the kidney a2A ~-adrenoceptors is indicated by the similar pKi-values obtained in our previous study where another approach was used (Uhl6n and Wikberg 1991 a). The drug pKi-values for a2-adrenoceptor subtypes determined in the present study do not represent agonist binding to high affinity forms of a2-adrenoceptors. We have previously shown that, with the spinal cord, the use of NaC1, Gpp(NH)p and EDTA totally eliminates the agonist high affinity binding sites of a2-adrenoceptors (Uhl6n and Wikberg 1991 b) . Moreover, the data for a number of other tissues indicate that these conditions will eliminate the high affinity agonist site of az-adrenoceptors (Michel et al. 1980; Snavely and Insel 1982) . Since the effects of Gpp(NH)p and EDTA are mediated via G-proteins and the effect of Na + is mediated by interaction with a specific aspartate residue (Horstman et al. 1991 ) which appears to be conserved among all G-protein coupled receptors, including all the a2-adrenoceptors cloned to date, it is conceivable that the inclusion of NaC1, Gpp(NH)p and EDTA will eliminate the agonist high affinity conformation for both U2A 1-and azAz-adrenoceptors. Since the results of the present study clearly indicate that the competition curves for strong agonists such as (-)-adrenaline and (-)-noradrenaline are fitted best into models that assume one site for the azA-adrenoceptor, it is clear that our assay conditions essentially eliminate the agonist high affinity state of the a2-adrenoceptors. Thus, the major difference in apparent affinities of guanoxabenz for aZA 1-and aZAz-adrenoceptors is not due to the formation of agonist high affinity states. All the binding sites studied here clearly represent a2-adrenoceptors because they show the expected stereoselective binding properties for (+)-and (-)-adrenaline and the affinities expected of classical a~-and az-adrenoceptor blockers such as yohimbine, rauwolscine, corynanthine and prazosin. Since dopamine showed much lower affinities than either (-)-adrenaline or (-)-noradrenaline, the sites cannot be classified as being dopamine receptors. The sites labelled do not represent imidazoline-binding sites ("I-receptors") since these invariably show low nonstereoselective affinities for catecholamines as well as a negligible affinity for RX821002 itself (Wikberg 1989; Wikberg and Uhl6n 1990; Wikberg et al. 1991; Langin et al. 1990) . In this context, it should be mentioned that the RG20-adrenoceptor was originally classified as an "a2Dadrenoceptor" (Lanier et al. 1991 ) because its pharmacology was similar to that of the rat submaxillary gland a2-adrenoceptor (Michel et al. 1989) , the latter being placed in a category termed "a2D" by Bylund et al. (1991) . However, we believe there are reasons to abandon the nomenclature a2D. Our data show clearly that the RG20 encodes a receptor with pharmacology closely similar to that of the other adrenoceptors studied in the rat which we have classified as being a2A-adrenoceptors. The original reason to choose the nomenclature a2A for these receptors was that among the first rat a2-adrenoceptors that we classified according to subtype was the cerebral cortex az-adrenoceptor which, in accordance with the original subtype classification of rat cerebral cortex a2-adrenoceptors by Bylund (1985) , was classified as the a2A-type (Uhl6n and Wikberg 1991 b) . Since all the other receptors investigated in the present study showed pharmacological properties virtually identical with those of the cerebral cortex a2A-adrenoceptor, it seemed quite logical to classify them all as being of the a2a-type. A comparison of the data of our present study (Table 1) with the data reported for the submaxillary gland "azi)adrenoceptors" (Michel et al. 1989; Bylund et al. 1991) reveals that the receptors show virtually identical phar-287 macological properties. The main reason that the rat submaxillary gland a2-adrenoceptor was placed in a category of its own seems to be that it shows a fairly low affinity for rauwolscine, yohimbine or WB4101 when compared with aa-adrenoceptors of human origin that had earlier been classified as being of the a2A-adrenoceptor type (see Bylund et al. 1991; Michel et al. 1989 ). However, in view of the fact that the RG20 gene is closely similar to the human a2-C10 gene, which encodes an az-adrenoceptor with the human azA-adrenoceptor profile (Kobilka et al. 1987; Harrison et al. 1991 a,b) , it seems quite reasonable that RG20 and a2-C10 are species variants of the same a2-adrenoceptor gene. We therefore presently prefer to view our azA-adrenoceptors as being variants in the rat corresponding to the human a2Aadrenoceptor. Recent data of Link et al. (1992) provide strong evidence that a single amino acid change from (Cys 201 in the human to Ser 2°t in rodent azA-adrenoceptots), is responsible for the low affinity of yohimbine for the rat azA-adrenoceptor. The present discussion prompts the need for an improved, general method for the classification of receptor subtypes; the best approach for the future will probably be based on the structure of the receptors. In summary, the most pertinent finding of the present study was that azA-adrenoceptors in rat tissues appear to be represented by two forms which we have termed azAt and a2A 2. The only drug which hitherto shows major selectivity for these two receptor forms is guanoxabenz. The present data, when combined with previous studies from our laboratory (Uhl6n and Wikberg 1991a; Uhl6n et al. 1992; Xia et al. 1993) , show that at least five forms of az-adrenoceptors are present in the rat: azAl, a2A2, a2B 1 and a2B 2 and a2c. Heterogeneity of alpha-2-adrenergic receptors Subtypes of a2-adrenoceptors: Pharmacological and molecular biological evidence converge Subtypes of a 1-and a2-adrenergic receptors Alpha-2A and alpha-2B adrenergic receptor subtypes: antagonist binding in tissues and cell lines containing only one subtype Pharmacological evidence for alpha-2C and alpha-2D adrenergic receptor subtypes Molecular cloning, sequencing and expression of an aa-adrenergic receptor complementary DNA from rat brain a) Molecular characterization of a 1-and a2-adrenoceptors b) Pharmacological characterization of rat a2-adrenergic receptors An aspartate conserved among G-protein receptors confers allosteric regulation of az-adrenergic receptors by sodium Cloning, sequencing, and expression of the gene coding for the human platelet a2-adrenergic receptor Binding of [3H]idazoxan and its methoxy derivative [3H]RX821002 in human fat cells: [3H]idazoxan but not [3H]RX821002 labels additional non-a2-adrenergic binding sites Isolation of rat genomic clones encoding subtypes of the a2-adrenergic receptor. Identification of a unique receptor subtype Cloning of two mouse genes encoding a2-adrenergic receptor subtypes and identification of a single amino acid in the mouse a2-C 10 homolog responsible for an interspecies variation in antagonist binding Protein measurement with the Folin phenol reagent Differences between the alpha 2-adrenoceptor in rat submaxillary gland and the alpha 2 A-and alpha 2B-adrenoceptor subtypes Differential regulation of the a2-adrenergic receptor by Na + and guanine nucleotides Characterization of a-adrenergic subtypes in the rat renal cortex. Differential regulation of a 1-and a2-adrenergic receptors by guanyl nucleotides and Na + Novel N-hydroxyguanidine derivatives as anticancer and antiviral agents Optimization of the Schiff bases of N-hydroxy-N'-aminoguanidine as anticancer and antiviml agents Delineation of three pharmacological subtypes of a2-adrenoceptors in the kidney Rat spinal cord a2-adrenoceptors are of the a2A-Subtype: Comparison with a2A-and a2B-adrenoceptors in rat spleen, cerebral cortex and kidney using 3H-RX821002 ligand binding Delineation of rat kidney a2A-and a2B-adrenoceptors with [3H]RX821002 radioligand binding: computer modelling reveals that guanfacine is an a2A-selective compound 3H]-MK912 binding delineates two a2-adrenoceptor subtypes in rat CNS one of which is identical with the cloned pA 2 d a2-adrenoce ptot The rat a2-C4 adrenergic receptor gene encodes a novel pharmacological subtype Design, synthesis testing, and quantitative structure-activity relationship analysis of substituted salicylaldehyde Schiff bases of 1-amino-3-hydroxyguanidine tosylate as new antiviral agents against coronavirus High affinity binding of idazoxan to a non-catecholaminergic binding site in the central nervous system: description of a putative idazoxan-receptor Further characterization of the guinea pig cerebral cortex idazoxan-receptor. Solubilization, distinction from the imidazole-site and demonstration of cirazoline as an idazoxanreceptor selective drug Medetomidine stereoisomers delineate two closely related subtypes of idazoxan (imidazoline) Ireceptors in the guinea pig Further evidence for the existence of two forms of a2n-adrenoceptors in the rat Molecular characterization of a rat a2B-adrenergic receptor Acknowledgements. We are indebted to Dr. Stephen M. Lanier, Department of Clinical Pharmacology, Medical University South Carolina, Charleston, S.C., for supplying us with the RG20-gene inserted into the pMT3 vector. The excellent technical support of Ms. Britt Jacobsson is greatfully acknowledged. This work was supported by the Swedish MRC (04X-05957), CFN (91-02), the Swedish National Board for Technical Development (89-02211P), and the Magnus Bergvall, Clas Groschinsky and ,~ke Wiberg foundations.