Identification of Multiple Binding Sites for Atrial Natriuretic Factor by Affinity Cross-linking in Cultured Endothelial Cells*

In a previous study, we found that atriopeptin I was much weaker > 500 nM) than atrial natriuretic factor (ANF-(8-33)) (ECB~ = 0.3 nM) at increasing cyclic GMP in cultured endothelial cells. In this study, we used the cross-linking reagent disuccinimidyl sub- erate to investigate whether the differences in activity were due to the presence of multiple ANF receptors. When 98% of the ANF-binding sites on endothelial cells were occupied by tyrosine-atriopeptin I after cross-linking, there was no difference in the concentration- response curve to ANF-(8-33) with regard to cyclic GMP accumulation. In contrast, when 96% of the bind- ing sites were occupied by cross-linked ANF-(8-33), a 60% decrease in the maximal cyclic GMP response was observed after the readdition of ANF-(8-33). These results suggest that ANF-(8-33) is binding to an additional site that atriopeptin I does not effectively bind. Affinity cross-linking of 12‘I-ANF to intact endothelial cells resulted in the labeling of two sites of M, -66,000 and -130,000. Approximately 94% of the ‘261-ANF binding sites had an M, -66,000. Labeling of this site was inhibited by both tyrosine-atriopeptin I (KI

In previous studies, we compared the binding of several atrial natriuretic peptides with their ability to increase cyclic GMP formation in bovine aortic endothelial (16) and smooth muscle (17) cells. Our results showed that the 21-amino acid peptide atriopeptin I was very effective at competing for lZ5I-ANF-binding sites, but was a very weak stimulator of cyclic GMP production (16, 17). In endothelial cells, 1 PM ANF-  increases cyclic GMP by s-fold, and at 10 nM an approximate 500-fold stimulation of cyclic GMP occurs in the presence of an inhibitor of phosphodiesterases. In contrast, atriopeptin I is a much less effective agonist. It begins to increase cyclic GMP at 10 nM and produces only a 7-fold increase in cyclic GMP at 0.1 p~. Because atriopeptin I was unable to antagonize the stimulation of cyclic GMP accumulation induced by the more potent atrial peptide ANF-(8-33), we proposed that endothelial cells may contain multiple ANF receptors (16). Typically, the identification of multiple receptors is established with selective agonists and antagonists for the specific receptor subtypes. However, no such selective probes are available for ANF receptors. In order to investigate further the possible existence of multiple ANF receptors in endothelial cells, we used affinity cross-linking techniques (18) in combination with cyclic GMP measurements. Our results with this novel approach provide definitive evidence for the existence of two ANF-binding sites of M, -66,000 and -130,000. Furthermore, our data indicate that only the less abundant site (M, -130,000) is coupled to guanylate cyclase and cyclic GMP formation. (HBSS) were purchased from Grand Island Biological Co. Calf serum was purchased from Sterile Systems Inc. Tissue culture dishes were from Falcon. Disuccinimidyl suberate was purchased from Pierce Chemical Co. Atriopeptin I, tyrosine-atriopeptin I, and atrial natriuretic factor-(8-33) were synthesized at Peninsula Laboratories, Inc. All other reagents were obtained as previously described (19,20).
Cell Culture-Endothelial cells were prepared and cloned from a bovine aorta as described by Longenecker et al. (21). The cells were maintained and subcultured in DMEM containing 10% calf serum, 584 pg/ml glutamine, 50 units/ml penicillin, and 50 pg/ml streptomycin. The cells were incubated at 37 "C in a 5% COZ humidified incubator.
Iodination of Atrial Peptides-In order to iodinate atriopeptin I, a tyrosine residue was added to the amino-terminal end of the native atriopeptin I. The tyrosine-atriopeptin I analog competed for "' Ityrosine-AP I and 12'I-ANF-(8-33)-binding sites and exhibited similar effects on cyclic GMP accumulation as the native atriopeptin I. Tyrosine-AP I and ANF-(8-33) (0.2 nmol) were incubated with 500-1000 pCi of NalZ5I for 15 min at room temperature in the presence of IODO-GEN (22). Unbound iodide was separated from the labeled peptide with a Sep-Pak ClS column (23). The specific radioactivity of the atrial peptides was 700-1400 Ci/mmol. Affinity Cross-linking Studies-Confluent endothelial cells in 35mm dishes were washed three times with 2 ml of HBSS containing 10 mM Hepes, pH 7.3. The cells were incubated in 1 ml of HBSS containing either 0.1 p~ tyrosine-AP I or ANF-(8-33) for 1 h at room temperature. One milliliter of HBSS containing 0.2 mM DSS (dissolved in dimethyl sulfoxide) was added, and the cells were incubated an additional 30 min at room temperature. The cells were washed four times with HBSS containing 2 mg/ml bovine serum albumin. The cross-linked cells were then used for either radioligand binding or cyclic GMP response studies.
Binding Studies-Affinity cross-linked or control endothelial cells were incubated in 1 ml of DMEM containing 25 mM Hepes, pH 7.3, 2 mg/ml bovine serum albumin, and either lZ5I-tyrosine-AP I or ' "1-ANF for 45 min at 37 "C. The cells were washed four times with HBSS containing 10 mM Hepes and 2 mg/ml bovine serum albumin. The cells were solubilized with 1 ml of 1 N sodium hydroxide, and the radioactivity was determined. Nonspecific binding was determined by the addition of 0.1 p~ unlabeled tyrosine-AP I or ANF-  to parallel cultures.
Cyclic GMP Determinations-Cross-linked or control cells (exposed only to DSS) were preincubated for 10 min at 37 "C with 990 pl of DMEM containing 25 mM Hepes, pH 7.3, and 0.5 mM isobutylmethylxanthine. ANF-(8-33) (10 pl) was added to the medium, and the cells were incubated for an additional 3 min. The medium which contained very little cyclic GMP was completely aspirated, and 750 p1 of 6% trichloroacetic acid was added to the cells. Trichloroacetic acid was removed from the cellular extracts with water-saturated ether. A 500-pl aliquot was acetylated (24), and the amount of the cyclic GMP was determined by radioimmunoassay as previouslv described (6, 16). SDS-Polvacrvlumide Gel EkctroDhoresis-Intact endothelial cell monolayers" weie incubated for 1 h with 1 nM lZ5I-ANF-(8-33) in HBSS containing 10 mM Hepes, pH 7.3, and increasing concentrations of either tyrosine-atriopeptin I or ANF-(8-33). DSS was added to a final concentration of 0.1 mM, and the cells were incubated for 30 min at room temperature. The medium was removed by aspiration, and 400 p1 of SDS sample buffer (62.5 mM Tris-HC1, pH 6.8, 10% glycerol (w/v), 5% 0-mercaptoethanol (v/v), and 2.3% SDS) was added to the cells. The culture plates were placed in boiling water for 3 min. The solubilized proteins were electrophoresed using a 7.5% separating gel according to Laemmli (25). The gels were dried and exposed to DuPont Cronex x-ray film at -70 "C to prepare autoradiograms which were then scanned with a densitometer. Some gels were sectioned and counted for radioactivity.

RESULTS
Intact endothelial cell monolayers were exposed to 0.1 PM tyrosine-atriopeptin I for 1 h. T h e bound tyrosine-atriopeptin I was cross-linked to ANF-binding sites by the addition of 0.1 mM DSS. To determine the efficiency of cross-linking, the cell cultures were then examined for their ability to bind lZ5Ityrosine-atriopeptin I.

I Binding of 1251-tyrosine-atrwpeptin I or 1251-ANF-(8-33) to bovine
aortic endothelial ceUs Confluent endothelial cells were incubated with 0.1 p~ tyrosineatriopeptin I or 0.1 p~ ANF-(8-33) for 1 h. In some cultures, the bound atrial peptide was cross-linked with 0.1 mM DSS, whereas control cultures received an equal volume of dimethyl sulfoxide. After 30 min, both the control and cross-linked cultures were extensively washed and then incubated with 250 p~ 1261-tyrosine-atriopeptin I (Experiment 1) or 200 PM lz'I-ANF-(8-33) (Experiment 2) for 45 min at 37 "C. Nonspecific binding in Experiment 1 was 0.64 f 0.016 fmol/ lo6 cells and 2.49 f 0.15 fmol/106 cells in Experiment 2. DSS refers to cells exposed to DSS only. Tyr-AP I -DSS and ANF-(8-33) -DSS refer to cultures pretreated with tyrosine-AP I or ANF-(8-33) and then exposed to dimethyl sulfoxide. Tyr-AP I + DSS and ANF-(8-33) + DSS refer to cultures pretreated with tyrosine-AP I or ANF ) and then exposed to DSS. Each value represents the mean of tridicate determinations f S.E. from a reoresentative exDeriment.

Experiment 1
Control Endothelial cells were also cross-linked with ANF-(8-33), which is a very potent stimulator of cyclic GMP accumulation. Table I (Experiment 2) demonstrates that cross-linking with ANF-(8-33) resulted in a large decrease in the subsequent binding of lZ5I-ANF- . In this case, approximately 96% of the binding sites were bound by cross-linked ANF-(8-33).  1 h, washed extensively, and then re-exposed to increasing concentrations of ANF-(8-33). We found that pretreatment of endothelial cells with ANF-(8-33) had no effect on the concentration-response curve with readdition of ANF-(8-33), suggesting that ANF-(8-33) does not promote the desensitization of guanylate cyclase under these conditions (data not shown).
To determine if endothelial cells possess multiple binding sites for ANF-(8-33) as suggested by the physiological studies, intact monolayers were cross-linked with '"I-ANF-  in the presence of increasing concentrations of either tyrosine-AP I or ANF-  and the cellular proteins were then separated by SDS-polyacrylamide gel electrophoresis. The autoradiographs show that l2'1-ANF-(8-33) labeled two binding sites (Fig. 2, A and B). The predominant binding site had a molecular size of approximately 66,000 daltons, whereas the second site labeled had a molecular size of approximately 130,000 daltons. By scanning autoradiographs with a densitometer, it was determined that the 130,000-dalton site represented 6.14 * 0.44% (S.D., n, = 5) of '*'II-ANF-(8-33)binding sites. Furthermore, when the bands were excised from the gel, it was determined that the 130,000-dalton band contained 6.6 _+ 0.56% (S.D., n = 5) of the radioactivity that was specifically incorporated into the 66,000-and 130,000-dalton binding sites (data not shown).
Tyrosine-atriopeptin I was compared to ANF-(8-33) for its ability to inhibit "'I-ANF-(8-33) labeling of the 66,000-and 130,000-dalton sites (Fig. 2, A and B ) . The degree of inhibition was quantitated by scanning the autoradiographs shown in Fig. 2 ( A and B ) with a densitometer. Fig. 2 (C and D ) shows that both tyrosine-AP I and ANF-(8-33) effectively inhibit labeling of the 66,000-dalton site. For this site, the KI of tyrosine-AP I was approximately 10-fold higher than the KI of ANF-(8-33) ( Table 11). Fig. 2C shows that 0.1 FM tyrosine-AP I inhibited labeling of the 66,000-dalton site to nearly the same extent as 0.1 p~ ANF- . However, at this concentration, tyrosine-AP I produced only a 4-fold increase in cyclic GMP compared to more than a 400-fold increase with ANF-(8-33) (Fig. 2, C and D). These results demonstrate that ANF binding to the 66,000-dalton site is not associated with an increase in cyclic GMP formation. Fig. 2A shows that tyrosine-AP I was less effective at inhibiting the labeling of the 130,000-dalton site when compared to the 66,000-dalton site and much weaker than ANF-(8-33) a t inhibiting labeling of the 130,000-dalton site. At this site, the Kr for tyrosine-AP I was at least 150-fold greater than the KI for ANF-(8-33) ( Table 11). In contrast to tyrosine-AP I, the KI for ANF-(8-33) at the 130,000-dalton site was similar to its KI for the 66,000-dalton site. Furthermore, the ECso for cyclic GMP stimulation was similar to the KI of ANF-(8-33) for these two sites. These results demonstrate that the 66,000-dalton site has a greater affinity for tyrosine-AP I than the 130,000-dalton site and that both of these sites have a similar affinity for ANF- .

DISCUSSION
In previous studies, we found that atriopeptin I was very effective at competing for 12'I-ANF-binding sites when compared to ANF-(8-33), but was a weak stimulator of cyclic GMP production in endothelial (16) and smooth muscle (17) cells. Several possible explanations may account for the apparent disparity between the binding and cyclic GMP responses. Assuming that these cells possess a single receptor for ANF, as suggested by a linear Scatchard plot (16, 17), these results could be due to differences in binding affinity for the two peptides. This possibility, however, is unlikely since the KI values for AP I and ANF-(8-33) were similar when the binding of '*'I-ANF-(8-33) was examined (16, 17). In endothelial and smooth muscle cells, the KI for AP I was only 6-fold higher than the KI for ANF- (8- The cells were solubilized with SDS sample buffer containing 8-mercaptoethanol and subjected to SDS-polyacrylamide gel electrophoresis on a 7.5% gel, and autoradiograms were exposed for 9 days. A similar Kl value for the 66,000-dalton band was obtained in an autoradiogram exposed for 3 days (data not shown). The mobilities of molecular weight standards are designated. C, densitometric scanning of the 66,000-dalton ( 0 ) and 130,000-dalton  N P ) , the amount of hound ""II-ANF-(8-33) present in the 66,000-and 130,000-dalton hand from parallel gels was 8.6 and 0.7 fmol/106 cells, respectively. The amount of radioactivity present in the bands was linear with the area of the peak obtained from densitometry, which was used to quantitate the relative optical density. Each determination for cyclic GMP accumulation represents the mean of triplicate dishes.

Cornparkon of KI and EC, values for tyrosine-A P I and A NF-(8-3.7)
The K I values were calculated using the equation Kl = ICm/(l + L/Kn), where Kl is the inhibition dissociation constant, ICso is the concentration required to inhibit 50% of ""I-ANF-(8-33) binding (obtained from Fig. 2, C and D), L is the concentration of "'1-ANF- for cyclic GMP stimulation that was 100-1500-fold higher for AP I than ANF-(8-33) (16,17). Furthermore, endothelial cells cross-linked with ANF-(8-33) had a 10-fold higher basal cyclic GMP level than those cells cross-linked with tyrosine-AP I.
These results suggest that ANF-(8-33) is much more effective than tyrosine-AP I in activating guanylate cyclase when both are bound to ANF-binding sites. A second possible explanation for these findings is that AP I readily binds to a single class of ANF receptors, but does not activate guanylate cyclase as effectively as ANF-(8-33) (i.e. the coupling to guanylate cyclase activation is different and ineffective). In order to address this possibility, we crosslinked cells with both tyrosine-AP I and ANF-(8-33) and then measured the cyclic GMP response to ANF- . Our results demonstrated that, when 98% of the binding sites were occupied by tyrosine-AP I, there was no alterat.ion in the concentration-response curve to ANF- .
These results could occur if there was a large number of spare receptors so that only 2% or less of the total ANFbinding sites are required for a maximal cyclic GMP response. If spare receptors were responsible for the inability of tyrosine-AP I to block the stimulation by ANF-(8-33), we would expect that cross-linking with ANF-(8-33) would produce a similar result. However, when cells were cross-linked with ANF-(8-33), there was a 60% decrease in the maximal cyclic GMP response. This observation suggests that ANF-(8-33) is binding to a second site that tyrosine-AP I does not effectively bind and that this site is coupled to guanylate cyclase activation and increased cyclic GMP formation. It seems apparent that, when the cells are cross-linked with tyrosine-AP I, many of the guanylate cyclase-coupled sites remain vacant, allowing ANF-  to bind effectively and to activate guanylate cyclase. In contrast, when the cells are cross-linked with ANF-(8-33), both the guanylate cyclase-coupled and -uncoupled sites are apparently occupied, leading to an increased basal level of cyclic GMP and a diminished maximal cyclic GMP response after the readdition of ANF-(8-33).
The existence of multiple ANF-binding sites was confirmed by SDS-polyacrylamide gel electrophoresis. Approximately 94% of the sites labeled by lz5I-ANF-(8-33) had an M, -66,000. Labeling of this site was decreased by both tyrosine-AP I and ANF- . The observation that the KI for tyrosine-AP I was approximately 10-fold higher than the KI for ANF-(8-33) is consistent with our previous competition binding studies with endothelial cells (16). Our finding that tyrosine-AP I effectively competes with lZ5I-ANF-(8-33) for the 66,000-dalton binding sites to a similar extent as ANF-(8-33) but does not elicit the large increases in cyclic GMP accumulation suggests that this site is probably not coupled to guanylate cyclase and cyclic GMP formation. If the 66,000dalton site mediated the increase in cyclic GMP, but AP I was an ineffective agonist at this site, we would have expected it to antagonize the ANF-(8-33)-induced stimulation of cyclic GMP. '"I-ANF-(8-33) also labeled a second site of M , -130,000, which comprised approximately 6% of the lZ5I-ANF-(8-33)-binding sites. With respect to this binding site, ANF-(8-33) was much more effective than tyrosine-AP I at decreasing the labeling of lZ5I-ANF- . Whereas the KI for tyrosine-AP I was 10-fold higher than the KI for ANF-(8-33) at the 66,000-dalton site, it was over 150-fold higher at the 130,000-dalton site. These findings demonstrate that the 66,000-dalton site has a higher affinity for AP I than the 130,000-dalton site and suggest that AP I may be potentially useful as a relatively selective peptide for this site. These results also suggest that the COOH-terminal amino acids, phenylalanine-arginine-tyrosine, are necessary for effective binding to the 130,000-dalton site, but not to the 66,000dalton site. Furthermore, we found that 0.1 PM tyrosine-AP I inhibited the binding of lZ5I-ANF-(8-33) by 35%, but produced only a 4-fold rise in cyclic GMP. In contrast, ANF-(8-33) elicited an approximate 100-fold stimulation of cyclic GMP at a concentration required to inhibit the labeling of the 66,000-dalton site by 35%. This observation and our previous studies (16,17) showing that atriopeptin I1 and atriopeptin I11 increase cyclic GMP to nearly the same extent as ANF-  suggest that the carboxyl-terminal phenylalanine-arginine are the important residues for coupling the ANF receptor to the activation of guanylate cyclase.
Our studies also showed that the EC50 for cyclic GMP stimulation by tyrosine-AP I is over 250-fold higher than the EC,, for ANF-(8-33). These results demonstrate that the EC5, for cyclic GMP stimulation by tyrosine-AP I and ANF-(8-33) correlates better with the KI of these peptides for the 130,000-dalton site. We have also cross-linked endothelial cells with '251-tyrosine-atriopeptin I and found essentially all the radiolabel in the 66,000-dalton protein (data not shown). Taken together, these findings suggest that the 130,000-dalton site is more likely the ANF receptor coupled to guanylate cyclase. Recent work in our laboratory (26) has demonstrated that highly purified particulate guanylate cyclase (Mr -120,000) from rat lung exhibits high affinity and specific binding for ANF-(8-33). These results suggest that the 130,000-dalton binding site in endothelial cells that is coupled to cyclic GMP formation may be particulate guanylate cyclase. However, additional studies are needed to determine the precise relationship between the 130,000-dalton site and guanylate cyclase.
Affinity cross-linking techniques have been also used to identify an ANF-binding site with an M, -130,000 in membranes from rat kidney (27) and bovine adrenal cortex (28, 29). In rabbit aorta membranes, three binding sites for ANF have been identified with molecular sizes of 60,000, 70,000, and 120,000 daltons (30). These differences may reflect mul-tiple ANF receptors or may result from the cellular heterogeneity of the tissue and species examined. From our results, we can conclude that cellular heterogeneity cannot account for the observation of multiple binding sites since cloned endothelial cells were used in these studies. Another advantage of using cloned, intact cells was that we were able to measure a physiological response (cyclic GMP synthesis) to ANF-(8-33) after cross-linking. The use of cross-linking studies coupled with cyclic GMP formation has also allowed us to examine the heterogeneity of ANF-binding sites when specific antagonists are unavailable. By taking this novel approach, we were able to identify multiple ANF-binding sites in endothelial cells and to determine that the binding site with M, -130,000 is most likely coupled to guanylate cyclase and cyclic GMP formation. We have also obtained very similar results with cultured bovine smooth muscle and bovine adrenal cortical cells (data not shown). We do not know the relationship of the 66,000-dalton binding site to the 130,000-binding site or its intracellular messenger. However, the approach used in these studies should be useful in identifying its intracellular messenger and possibly the messengers coupled to other hormone receptor systems. In order to facilitate future discussion of these two ANF-binding sites, we suggest that the guanylate cyclase-coupled site (M, -130,000) be designated ANF-R1 and the uncoupled site (M, -66,000) ANF-R2.