Dynamics of Atrial Natriuretic Factor-Guanylate Cyclase Receptors and Receptor-Ligand Complexes in Cultured Glomerular Mesangial and Renomedullary Interstitial Cells*

The dynamics of the guanylate cyclase receptor of atrial natriuretic factor (GCA-ANF receptor) were investigated in cultured glomerular mesangial and re- nomedullary interstitial cells from the rat. In these cells, the GCA-ANF receptor did not mediate internal- ization and lysosomal hydrolysis of 1251-ANF1-2s and did not undergo ligand-induced endocytosis. Glomeru- lar mesangial cells were able, however, to mediate internalization and lysosomal hydrolysis of '261-ANF1-28 via clearance ANF (C-ANF) receptors and to promote rapid receptor-mediated internalization and lysosomal hydrolysis of 12'I-(Sar') angiotensin 11. Radioligand specifically bound to surface GCA-ANF receptors was rapidly dissociated at 37 "C (& > 0.8 min"), with a Q10(30-37 "C) > 6. The dissociation was markedly slower at subphysiological temperatures (Qlo(4- 30 "C), 2-3) or in the presence of 0.5 mM amiloride. The results demonstrate that the GCA-ANF receptor, contrary to C-ANF receptors and most other polypeptide hormone receptors, is a membrane resident protein that does not mediate internalization and lysosomal hydrolysis of ligand. The termination of the interaction of ANF with GCA-ANF receptors results from a physiological process that leads to rapid dissociation of receptor-ligand complexes. The unique dynamics of GCA-ANF receptor-ligand complexes are likely to con- tribute importantly to stimulus-response homeostasis of ANF. The re-spectively; Nonspecific binding (1251-ANFl.28 bound in the presence of 0.3 p~ unlabeled ANF1-zR) was subtracted from total binding, and a 1:1 stoichiometry for receptor-ligand interaction was assumed to calcu- late the concentration of receptors.

biological receptor proper (originally named type B, or type I (Rl), and more recently, guanylate cyclase (GC) receptor) is a 120-130-kDa protein that contains in its cytoplasmic domain guanylate cyclase-and tyrosine kinase-like sequences (1). Two subtypes of GC receptors, named GCA and GCB, have been identified (2,3). GCB receptors are unlikely to be physiological receptors of ANF because they display a very low affinity for the native hormone but have a relatively high affinity for CNP, another member of the natriuretic peptide family (4). GCA-ANF receptors mediate most if not all functional effects of ANF and have as their main known second messenger cGMP (5)(6)(7)(8). The other major class of ANF receptors is named clearance (C) or type I1 (R2) receptor. The C-ANF receptor is a homodimer of 120-130 kDa (60 kDa under reducing conditions) which has a very short cytoplasmic domain, a property that is shared by other clearance or transport receptors (8)(9)(10)(11). C-ANF receptors do not mediate any of the known end-organ effects of the hormone in renal and vascular tissues (e.g. natriuresis, vasorelaxation) but have an important role in the removal of ANF from the circulation (12)(13)(14)(15)(16). C-ANF receptors are by far the most abundant class of ANF receptors, comprising more than 95% of the total ANF receptor population in kidney cortex and vascular tissues (14,15,17).
Receptor-mediated endocytosis of ligand is a common property of polypeptide hormone receptors which contributes to ligand metabolism, termination of receptor-ligand interactions, and, in many instances, to the regulation of receptor cell surface density (18,19). We have shown previously that C-ANF receptors exert their clearance function by receptorligand internalization, lysosomal hydrolysis of endocytosed ANF, and recycling of internalized C-ANF receptors to the cell surface (15). Very little is known, however, regarding the dynamics of membrane GC-ANF receptors and of GC-ANF receptor-ligand complexes. In the present study we determined these processes in cultured glomerular mesangial and renomedullary interstitial cells from the rat. The results will show that the GCA-ANF receptor is a membrane resident protein that, contrary to C-ANF receptors, and most other polypeptide hormone receptors, undergoes minimal, if any, endocytosis. Consequently, GCA-ANF receptors do not mediate internalization and lysosomal hydrolysis of ANF. The data will further show that the interaction of ANF with GCA-ANF receptors is rapidly terminated by a process that leads to a marked increase in the dissociation of receptor-ligand complexes a t physiological temperatures. ANF1., 8 (ANFWl26) and des[Gln", Ser", Gly'", Leuz1, Gly22]ANF4~z~-NH~ (C-ANF,.,,) were generous gifts from Dr. John Lewicki, California Biotechnology Inc., Mountain View, CA. The following materials and substances were purchased commercially: culture flasks and six-well culture plates (Falcon-Becton Dickinson Co., Lincoln Park, NJ), defined fetal bovine serum (HyClone Laboratories Inc., Logan, UT), carrier-free NaIz5I (Amersham Corp.), HEPES (Research Organics Inc., Cleveland, OH), acetonitrile and trifluoroacetic acid (Pierce Chemical Co.). (Sar')-angiotensin I1 (Sar'-ATII), Dulbecco's modified Eagle's medium, RPMI 1640 medium, Hanks' balanced salt solution, bovine serum albumin, amiloride, and all other reagents were obtained from Sigma.

Materials--Rat
Culture Media and Binding Solution-The composition of culture media and binding solutions was as follows. Solution A was Hanks' balanced salt solution supplemented with 3.6 g/liter HEPES, 100 units/ml penicillin, 0.1 mg/ml streptomycin, and 0.25 pg/ml amphotericin B, pH 7.4. Solution B was RPMI 1640 supplemented with 20% fetal bovine serum, 2 g/liter NaHC03, 3.6 g/liter HEPES, 0.6 unit/ml insulin, 100 units/ml penicillin, 0.1 mg/ml streptomycin, and 0.25 pg/ml amphotericin B. Solution C was Dulbecco's modified Eagles's medium (supplemented with 10% fetal bovine serum, 3.7 g/ liter NaHC03, HEPES, 0.6 unit/ml insulin, and the antibiotics and antifungal agents indicated above) conditioned with 3T3 Swiss Albino mouse fibroblasts (ATCC) during their log phase of growth. Binding solution was RPMI 1640 supplemented with HEPES (3.6 g/liter) to a pH of 7.4, bovine serum albumin (2 mg/ml), and insulin (0.6 unit/ ml).
Cell Cultures-A homogeneous ("cloned) cell line of glomerular mesangial cells was obtained from rat glomeruli primary culture by minor modifications of the procedure of Harper et al. (22). Briefly, isolated glomeruli were prepared by sequential sieving from dissected kidney cortex of five Sprague-Dawley rats weighing 50-70 g. The resulting glomeruli were suspended in 25-cm2 plastic bottles with solution A and digested with solution A containing 2.5% (w/v) trypsin for 15 min at 37 "C. Trypsinized glomeruli were washed once with solution A and were further digested with solution A containing 200 units/ml collagenase type I for 40 min at 37 "C. The cells were plated in culture flasks and grown in a 1:l mixture of solutions B and C. After 2 weeks the cultures were constituted predominantly by nonepithelial cells. At this point the cells were split and plated at low density (about 100 cells/ml) in 100-mm-diameter dishes in a 1:l mixture of solutions B and C. Individual colonies were harvested using cloning cylinders. A single expanded clone with the morphological features of mesangial cells was used for all the experiments. Cultured glomerular mesangial (GM) cells were used either in early (-40th) passages (EGM) or late (>30th) passages (LGM). Renal medullary interstitial (RMI) cells were prepared as described previously (23) and studied in the 11th-15th passages from the primary culture. Cultures were maintained in a C02 incubator (5% CO,, 95% atmosphere gas) at 37 "C. EGM, LGM, and RMI cells were routinely subcultured at 1/10-1/20 dilution after dissociation with Ca2+-Mp2+free Hanks' balanced salt solution medium supplemented with 0.025% trypsin, 0.5 mM EDTA, and 20 units/ml collagenase. When plated in six-well dishes, EGM and RMI cells reached confluence in approximately 7-10 days, whereas LGM cells reached confluence in approximately 3-4 days. In each experiment with confluent monolayers in six-well dishes, two to three wells were used for cell counts. For this purpose the monolayers were dissociated as described above, and the cells were counted in a hemocytometer.
Internalization, Metabolism, and Dissociation of Specifically Bound 125 I-ANFl-2~-The fate of specifically bound 1251-ANFl.2, was determined by chase experiments, as described previously (15). Briefly, cultured EGM, LGM or RMI cells grown to confluence in six-well dishes, were washed three times with 2 ml of ice-cold serum-free bicarbonate-free binding solution. Then, cells were incubated for 2 h at 4 "C with 1 ml of binding solution containing 1251-ANFl_28, diluted with unlabeled ANF1.28 to a specific activity of approximately 500 Ci/ mmol, to give a saturating concentration of 2 nM (15, 23). After washing with ice-cold binding solution to remove unbound ligand, cells were further incubated at 37 "C with fresh binding solution in the absence of added radioligand for several time intervals. In some experiments in LGM and RMI cells, 0.3 p~ C-ANF,-,, was added to the medium to prevent binding of radioligand to the small, if any, amount of C-ANF receptors present in these cells. The results of these experiments in LGM and RMI cells did not differ from those in which C-ANF,.,, was not added. Experiments were performed in the presence or absence of the lysosomotropic weak bases NH4Cl(10 mM) or chloroquine (0.1 mM). At these concentrations, the lysosomotropic agents were previously shown to block lysosomal hydrolysis of internalized 1251-ANFl.28 in BAVSM cells (15). In separate experiments, amiloride was added to the medium to a final concentration of 0.5 mM, 5 min before the incubation at 37 "C, to determine its effect on the fate of bound ligand.
The temperature dependence of the dissociation of surface GC-ANF receptor-ligand complexes was determined in LGM cells preloaded with saturating concentrations of 1251-ANFl.28 (2 nM) at 4 "C, as described above. After washing with ice-cold binding solution, cell monolayers were incubated further for time intervals ranging from 0.5 to 60 min at 4, 10, 20, 30, or 37 "C in fresh binding medium containing 1 p~ unlabeled ANFl.28 to preclude rebinding of the radioligand.
Determination of Radioactiuity at the Cell Surface, in Intracellular Compartments, and in Medium-At the end of incubation, cells were washed three times with 2 ml of ice-cold solution A. Bound ligand was removed from the cell surface by incubating the cells with an hypertonic acid solution (0.2 M acetic acid in 0.5 M NaC1) at 4 "C for 15 min. Intracellular radioactivity was determined in acid-washed cells. For this purpose, after the acid wash, cell monolayers were dissolved by incubation with a 2 N NaOH solution for 30 min at 37 "C. For the determination of intact and degraded ligand in medium, the incubation medium was mixed with ice-cold trichloroacetic acid/ phosphotungstic acid solution to a final concentration 6%/l% (w/v), and precipitate (intact ligand) and supernatant (degraded ligand) were separated by centrifugation. Radioactivity counts were determined in a y-spectrometer. Nonspecific binding, receptor-independent internalization, and receptor-independent hydrolysis of radioligand were determined in parallel, by adding a 150-fold molar excess of unlabeled ligand (0.3 p~) .
Results for each time of incubation were normalized in relation to the initial specific surface binding of radioligand at 4 "C. The nature of the radioactivity in medium and in acid wash (membrane bound) was also analyzed by reverse-phase HPLC as described previously (21).
Trypsinization of Surface Receptors and Solubilization of Celk-Total cell and intracellular concentrations of receptors in LGM cells were determined by a trypsinization-solubilization protocol, as described previously (15). Cell monolayers were preloaded with excess ANF,-,, (0.1 p~) at 4 "C for 2 h. After washing, cell monolayers were warmed to 37 "C in fresh medium for several time intervals to elicit internalization of receptors. Then, monolayers were washed three times with Ca2+-MgZ+-free solution A and incubated for 30 min at 4 "C with 0.5 ml of the same solution containing 5 mM EDTA and 1 mg/ml trypsin. At the end of incubation, the trypsin effect was neutralized by adding soybean trypsin inhibitor and calf serum to final concentrations of 1 mg/ml and lo%, respectively. Control cells were treated in an identical manner, except that trypsin was not added to the incubation mixture.
To determine the degree by which the trypsin treatment removed specific surface binding sites of ANF, trypsinized and control cells were incubated for 4 h at 4 "C with saturating concentration of ANF,-,, (2 nM), in presence or absence of trypsin, as described above. It was found that the trypsin treatment for 30 min removed 90-95% of the specific surface binding sites for lz5I-ANF,-28. Prolongation of the incubation with trypsin to 1 h, or increase in the trypsin concentration to 2.5 mg/ml, removed > 99% of surface-specific binding sites of ANF but resulted in some detachment of cells from the monolayers. Consequently, the 30-min incubation time with 1 mg/ml trypsin was chosen.
After trypsinization, cells were scrapped from the wells and washed three times with solution A containing 10% calf serum. Trypsinized and control cells were solubilized by a slight modification of the method of Marshall et al. (24). Briefly, the cells were suspended and incubated for 1 h at 4 "C in 500 pl of Ca*+-M?-free Hanks' balanced salt solution containing 35 mM HEPES (pH 7.4), 0.4% octaethyleneglycol dodecyl ether (C1208), 1 mM EDTA, 1 mM EGTA, 1 mg/ml soybean trypsin inhibitor, 20 pg/ml leupeptin, 1 mg/ml bacitracin, and 0.5 mM phenylmethylsulfonyl fluoride. At the end of incubation, 50 pI of 2% bovine y-globulin and 550 pl of 30% w/v polyethylene glycol (molecular weight, 8,000) were added to the suspension. The cells were centrifuged, and after washing with binding solution, the pellet was reconstituted in Hanks' balanced salt solution (supplemented with 35 mM HEPES to a pH of 7.4) containing0.025% (21208, 0.2% bovine serum albumin, 0.6 mg/ml egg white lysozyme, and soybean trypsin inhibitor, leupeptin, bacitracin, and phenylmethylsulfonyl fluoride at the above indicated concentrations. The concentration of solubilized ANF receptors was determined as reported previously (15). Briefly, 80 p1 of the solubilized receptor solution was mixed with 50 pl of binding solution containing ANFl-za and unlabeled ANF1.** to a final concentration of 2 nM. Equilibrium was allowed to proceed in microcentrifuge tubes for 24 h a t 4 "C. Receptor-ligand complexes were precipitated with 50 p1 of 0.2% bovine y-globulin and 250 pl of 34% w/v polyethylene glycol, and the microtubes were centrifuged for 5 min at 12,000 X g. Supernatants were aspirated, and the pellets were counted for radioactivity. Nonspecific binding (1251-ANFl.28 bound in the presence of 0.3 p~ unlabeled ANF1-zR) was subtracted from total binding, and a 1:1 stoichiometry for receptor-ligand interaction was assumed to calculate the concentration of receptors.

RESULTS
The values for the density (Bmax) and equilibrium dissociation constant ( k d ) a t 4 "C of the total surface ANF receptor population in LGM, EGM, and RMI cells used in the present experiments are summarized in Table I. This table also gives the relative distribution of ANF receptor subtypes, and the ANFl-28-induced generation of cGMP in these cell lines. EGM, LGM, and RMI cells have 8,000-30,000 ANF receptors/cell. Competition binding studies using C-ANF4-23 or saturation binding experiments with '251-(Y3)-C-ANF3-23, specific ligands of C-ANF receptors which do not bind to guanylate cyclase receptors (12, 17, 23), revealed that > 85% of the total population of ANF receptors in LGM and RMI cells is constituted by GC-ANF receptors. Cross-linking experiments in LGM cells with '251-ANF1-2s revealed a single autoradiographic band that migrated at approximately 130 kDa in sodium dodecyl sulfate-polyacrylamide gel electrophoresis under reducing conditions (not shown), confirming that LGM cells express GC-ANF receptors and only a minimal amount, if any, of C-ANF receptors. The very high affinity of these receptors for ANF1-28 at 4 "C ( k d < 0.1 nM) indicates that they belong to the GCA, rather than the GCB, subtype (4). For unknown reasons EGM cells, albeit derived from the same clone as LGM cells, also express C-ANF receptors in variable proportions (10-40% of the total ANF receptor population). The particular clone of EGM cells used in the experiments to be presented below expressed GCA-ANF and C-ANF receptors in a 6040 relationship, respectively. Table I further shows  that ANFl-z8 markedly increased cGMP levels in EGM, LGM, and RMI cells, with an ED60 of < 1 nM. The maximal absolute increase in cGMP in these cells was much higher, and the ED50 of the effect was much lower than those previously reported to occur in cultured BAVSM cells, a cell type that has only a very small proportion of guanylate cyclase receptors (7, 15). Fig. 1 depicts data on the fate of 1251-ANF1-28 initially bound to surface GCA-ANF receptors in LGM cells. Panel A shows that specifically bound radioactivity disappeared rapidly from the cell surface with a half-time of < 3 min when cells were warmed from 4 to 37 "C. However, this disappearance was accompanied by only minimal (<10%/60 min) internalization of radioactivity (panel B ) or hydrolysis of the radioligand (panel C). Practically all radioactivity that disappeared from the cell surface appeared as intact 1251-ANF1-28 (as assessed by trichloroacetic acid precipitability) in the medium, with a time course that precisely matched its disappearance from the cell surface (panel D). HPLC analysis of the medium demonstrated that the bulk of the radioactivity co-eluted with intact '251-ANF1-28 and that minimal amounts eluted as labeled metabolites (not shown). Fig. 1 further shows that NH4Cl, a lysosomotropic weak base that blocks lysosomal hydrolysis of ANF (15), had no detectable effects on binding, dissociation, or hydrolysis of '251-ANF1-2s in LGM cells.
The results above demonstrate that the disappearance of specifically bound radioligand from the cell surface of LGM cells is caused by a rapid dissociation of 1251-ANF1_28 from GCA-ANF receptors when cells are warmed from 4 to 37 "C. Because amiloride has been previously shown to increase the affinity of GC-ANF receptors in membranes from bovine adrenal glomerulosa (25), we tested the effects of this compound on the dynamic of GCA-ANF receptor-radioligand complexes in LGM cells. The results of these experiments are also shown in Fig. 1 (diamond symbols). As can be seen, amiloride markedly decreased the disappearance rate of radioligand from the cell surface by more than %fold (panel A ) and correspondingly decreased the rate of appearance of radioligand in the medium (panel D). The half-time of residence of specifically bound lZ5I-ANF1-28 at the membrane surface of amiloride-treated cells was approximately 9 min, whereas it was < 3 min in controls (panel A ) . In spite of this marked increase in resident time, there was no evidence for internalization (panel B ) and for NH4C1-sensitive (lysosomal) hy-  constant (k,,) of the total population of ANF receptors were determined in confluent EGM, LGM, and RMI cell monolayers by saturation binding experiments at 4 "C with 1251-ANFl~,, (see "Experimental Procedures"). Bmax and kd values for RMI cells between the 10th and 17th passages were taken from Fontoura et al. (23). The proportion of C-ANF receptors (in percent of total population of ANF receptors) was determined by competition binding experiments at 4 "C between 1251-ANF,.z, and saturating concentrations (0.3-1 phi) of C-ANF4.23, a specific ligand of C-ANF receptors, as reported previously (15, 23).
When LGM cells were preloaded with trace concentrations (10 PM) of '251-ANF1-2s instead of saturating concentrations of radioligand, the results were indistinguishable from those depicted in Fig. 1 (not shown). The HPLC profiles shown in Fig. 2 demonstrate that there was only a minimal appearance of hydrolytic products in the medium when LGM cells were incubated for 2 h at 37 "C in the continuous presence of 20 pM "sI-ANFl-,s with or without 10 mM NH&l or 0.1 p M unlabeled ANF1-28. Fig. 3 demonstrates that the dynamics of receptor-radioligand complexes in RMI cells, a cell type in which the overwhelming majority of ANF receptors is also constituted by  (Table I), is indistinguishable from that observed in LGM cells.
The above results show that at physiological temperatures the offset of radioligand from GCA-ANF receptors is very fast. The temperature dependence of the off-rate of 1251-ANF1_28 from GCA-ANF receptors in LGM cells is summarized in Table 11. In these experiments the kOff was determined from the initial monoexponential release of specifically bound radioligand to the medium which, at the higher temperatures (30,37 "C), accounted > 85% of the dissociation of lZ5I-ANF1-28 from GCA-ANF receptors. At lower temperatures, a second sluggish exponential component, which accounted for > 50% of radioligand dissociation, was too slow to be determined with accuracy in the time (1 h) of the experiments. The initial kOff at 4 "C (0.015 min") is very slow, consistent with the very low equilibrium dissociation constant ( k d ) at this temperature ( Table I). As expected, koff increases with temperature (Qio, 2-3) from 4 to 30 "C. Between the temperatures of 30 and 37 "C, however, there is an unexpected disproportionate 4fold increase in the koff from 0.18 to 0.85 min" (Qlo > 6). At 37 "C, 51% of specifically bound radioligand had already been released to the medium by 0.5 min, the first sampling time. Therefore, 0.85 min" represents a minimal value for the dissociation rate constant at 37 "C. The disproportionate increase in the off-rate at 37 "C, but not at subphysiological temperatures, indicates that this phenomenon depends on the metabolic integrity of the cells, probably on the generation of ATP (see "Discussion").
Chase experiments with 1251-ANF1_2s were also performed in EGM cells, a cell type that was derived from the same clone as the LGM cells (see "Experimental Procedures") but has a significant proportion (40%) of C-ANF receptors in addition to GCA-ANF receptors (Table I). Fig. 4 shows the results of these experiments. The decay of specifically bound radioligand from the cell surface (panel A ) is clearly bimodal, with a fast component similar to that observed in LGM and RMI cells (Figs. 1 and 3) and a slower component that is similar to that reported previously for C-ANF receptors in BAVSM cells (15). The appearance of trichloroacetic acidprecipitable radioactivity in the medium (panel D ) has a single component that corresponds precisely to the fast component of radioligand decay from the cell surface. This component is similar to that observed in LGM and RMI cells ( Figs. 1 and 3) and sharply contrasts with that observed previously in BAVSM cells, in which the appearance of trichloroacetic acid-precipitable radioactivity in medium is neg-

TABLE I1
Temperature dependence of the dissociation of '251-ANG1 2R from GCA-ANF receptors in cultured LGM cells LGM cell monolayers were preloaded with saturating concentrations of 12511-ANFl~,s (2 nM) at 4 "C as described under "Experimental Procedures." After washing with ice-cold binding solution to removed unbound ligand, cells were incubated in fresh binding medium containing 1 PM ANF1.2, and further incubated at 4, 10, 20, 30, or 37 "C for several time intervals from 30 s to 60 min. The loss of specifically bound radioligand from the cell surface and its appearance as intact '2511-ANFl.28 in medium were determined as described under "Experimental Procedures." The k,, was determined from the initial monoexponential release of specifically bound radioligand to the medium (see text). The Qlo values were calculated for temperature intervals of 4-10, 10-20,20-30, and 30-37 "C. We investigated further whether the lack of internalization and lysosomal hydrolysis of ANF in LGM was caused by a generalized defect of receptor-mediated endocytosis and/or lysosomal function in these cells. For this purpose we determined the dynamics of AT11 receptors using '251-Sar'-ATII as a radioligand and the same protocol as that used to investigate the dynamics of GCA-ANF receptors in these cells. LGM cells are able to promote fast receptor-mediated inter-nalization and lysosomal hydrolysis of 1251-Sar'-ATII, indicating that these cells do not have a defective endocytic-lysosomal apparatus.
The very rapid dissociation of '251-ANF1-28 when LGM or RMI cells are warmed from 4 to 37 "C precludes a conclusion on whether there is ligand-induced internalization of unoccupied GCA-ANF receptors in the experiments depicted in Figs. 1 and 3. T o investigate this issue we directly determined the time-dependent changes in the intracellular concentration of GCA-ANF receptors after exposure of LGM cells to saturating concentrations of ligand (see "Experimental Procedures"). Fig. 6 shows that after exposure to ANF, the total as well as the intracellular concentration of GCA-ANF receptors in LGM cells remain constant throughout the experiment. This result contrasts with that reported previously for BAVSM cells exposed to ANF, in which there is first an increase in the intracellular concentration of C-ANF receptors (internalization) and then a decrease toward control values (recycling) (15). The lack of change in the intracellular concentration of GCA-ANF receptors in LGM cells further demonstrates that ANF does not induce the internalization of occupied or unoccupied GCA-ANF receptors in these cells.

DISCUSSION
The present study unveils two major properties of the dynamics of membrane GCA-ANF receptors and of GCA-ANF

FIG. 6. Total cell and intracellular concentration of ANF receptors in LGM cells after exposure to ANF.
LGM cell monolayers were exposed to excess ANF1-28 (loe7 M) for 2 h at 4 "C. After removal of unbound ligand, cells were warmed and incubated at 37 "C for the indicated times to provide a stimulus for receptor internalization. At the end of each incubation time, one group of monolayers was incubated with trypsin as described under "Experimental Procedures." In separate experiments the trypsin treatment was found to remove 90-95% of specific cell surface binding sites of "'I-ANF1.28 (see "Experimental Procedures"). Another group of cell monolayers was subjected to the same incubation protocol, except that trypsin was not added to the incubation medium. ANF receptors from trypsinized and nontrypsinized monolayers were solubilized, and their concentration was determined by equilibrium binding with saturating concentrations (2 nM) of 1251-ANFl.28 at 4 'C, as described under "Experimental Procedures." Receptors obtained from nontrypsinized cells (filled circles) were taken to represent total (membrane receptor-ligand complexes in cultured renal cells. First, in these cells GCA-ANF receptors do not internalize at detectable rates and do not mediate internalization and subsequent lysosomal hydrolysis of ANF. Second, the termination of the interaction between GCA-ANF receptors and ANF at the cell surface results from a very rapid dissociation when the ligand binds to the receptor at physiological temperatures. The lack of internalization of GCA-ANF receptors is not caused by characteristics of a single specific cell type or by a generalized defect of the endocytic-lysosomal apparatus in the cultured renal cells used in the present experiments. Indeed, the dynamics of GCA-ANF receptors in two different species of cultured renal cells (LGM and RMI cells) were practically the same (Figs. 1 and 3). Furthermore, LGM cells, although failing to endocytose and hydrolyze ANF, conserve the ability to promote rapid receptor-mediated internalization and lysosomal hydrolysis of AT11 (Fig. 5). Finally, EGM cells, which express both GCA-ANF and C-ANF receptors in near equal proportions, handle ANF in a manner consistent with the presence of the two subtypes of receptors, namely internalization and lysosomal hydrolysis via C-ANF receptors and rapid off-rate via GCA-ANF receptors. It is noteworthy that the relative degree of internalization and offset of specifically bound radioligand in EGM cells is similar to the relative proportion of C-ANF and GCA-ANF receptors in these cells.
The dynamics of surface GCA-ANF receptors were followed by using its specific radioligand as a marker. In view of the very rapid off-rate of '251-ANF1_28 at 37 "C, it was possible that internalization of unoccupied GC-ANF receptors could have remained undetected. However, even when the off-rate of the radioligand was markedly decreased by amiloride, no significant internalization of receptor-ligand complexes or of lysosomal hydrolysis of '251-ANFl_2s was detectable (Fig. 1). I t could be argued that amiloride itself, perhaps by acidifying the cells, may have blocked internalization. The following additional evidence, however, strongly supports the conclusion that GCA-ANF receptors of LGM cells are not endocytosed (i) there is lack of significant hydrolysis of '251-ANF1-28 even when LGM cells are incubated in the continuous presence of saturating or trace amounts of radioligand for up to 2 h at 37 "C (Fig. 2); and (ii) ANF does not induce a timedependent change in intracellular concentration of GC-ANF receptors in LGM cells (Fig. 6).
Membrane receptors that are constitutively concentrated in coated pits or are clustered in these or other endocytic regions when complexed with their ligands are internalized at very fast rates, which in some instances may exceed 100% of surface receptors/min (16). Receptors for low density lipoprotein (19, 27), several other clearance and/or transport receptors (18, 2L-32), and polypeptide-hormone receptors such as insulin, epidermal growth factor (26,(33)(34)(35), and AT11 (Fig.  5) are examples of this category of receptors that undergo rapid endocytosis. Naturally occurring or experimentally induced mutations in which "internalization signals" in the cytoplasmic domain of these receptors are deleted or altered markedly reduce the endocytic rate to less than lO%/min but do not abolish endocytosis (26,27,29,32,35). In the case of the low density lipoprotein receptor, only the deletion of the entire cytoplasmic domain abolishes detectable internalization (27). It is generally assumed that relatively slow (<lo%/ min) but significant rates of endocytosis correspond to the entrapment of highly mobile membrane receptors in coated pits or other endocytic regions. On the other hand, rapid receptor-mediated endocytosis depends on specific biochemical interactions between the cytoplasmic domain of receptors with constitutive proteins of the endocytic apparatus (26).
We have demonstrated previously that C-ANF receptors in cultured BAVSM cells are constitutively internalized a t a relatively slow rate of 4-5%/min and that ligand binding does not increase this rate (15). This would suggest that C-ANF receptors are mobile membrane proteins that, with or without their ligand, are internalized as they become entrapped in endocytic regions. However, the present study demonstrates that endocytosis of GCA-ANF receptors in GM and RMI cells is minimal (<O.l%/min) even in presence of ligand. These dynamics, characteristic of bona fide membrane resident proteins (36), are highly uncommon for polypeptide-hormone receptors. The only other known example in this regard is the insulin receptor in IM-9 human lymphocyte cells, which, contrary to insulin receptors in other cell types, does undergo only minimal, if any, detectable internalization (33,34).
The present experiments do not exclude the possibility that GCA-ANF receptors may internalize in cells other than GM and RMI cells. Thus, it has been reported that GC-ANF receptors of Leydig tumor and pheochromocytoma (PC12) cells mediate internalization and lysosomal hydrolysis of ANF (37,38). In these studies the degree of constitutive internalization of GC-ANF receptors was not determined. It is possible, even likely, that tumor cells behave abnormally with respect to the dynamics of receptors and receptor-ligand complexes. Ligand-induced internalization of most biological receptors is accompanied by a prolonged decrease in cell surface receptor density (down-regulation). When receptors fail to internalize, as exemplified by insulin receptors in human lymphocyte IM-9 cells, homologous (ligand-induced) down-regulation does not occur (34). The reported lack of detectable in vivo regulation of GCA-ANF receptors in glomeruli and renal papillae of the rat, when plasma levels of ANF are altered by salt diet (39,40), is consistent with the present evidence that these receptors do not internalize at appreciable rates. Further experiments are needed, however, to test whether GCA-ANF receptors in intact or freshly isolated cells and tissues have the same dynamics as those observed in cultured LGM and RMI cells.
Known molecular features of GCA-ANF receptors are also consistent with its characteristics as a membrane resident protein. The binding domain of GCA-ANF receptors and the enzyme that generates the second messenger (guanylate cyclase) are part of a same molecule (1). The "mobile receptor theory," which states that ligand effects depend on the coupling of different molecular species, a mobile acceptor (the receptor proper) and an enzyme effector (41), is not applicable t o GCA-ANF receptors. Thus, mobility of GCA-ANF receptors is, theoretically, not necessary for coupling ANF binding to activation of guanylate cyclase. Moreover, the cytoplasmic domain of GCA-ANF receptors does not contain sequences of amino acids that have been associated with the internalization of many other receptors, including the NPXY/F (where X is any amino acid) sequence, nor does it contain an aromatic amino acid (tyrosine or phenylalanine) near the transmembrane domain (26,27,29,32,35). The nearest aromatic amino acid in the cytoplasmic domain of GCA-ANF receptors is a tyrosine, separated from the transmembrane domain by 42 amino acids (1). As pointed out above, however, deletion or substitution of internalization signals reduces but does not preclude internalization. Thus, it is unlikely that the absence of internalization signals that confer a high endocytic rate account entirely for the behavior of GCA-ANF receptors as a membrane resident protein. Further studies are needed to elucidate this question.
Internalization of receptor-ligand complexes may determine, at least in part, the termination of biological effects of polypeptide hormones. In the absence of internalization, other processes must be invoked to explain the termination of GCA-ANF receptor-mediated responses. Several complex ligandinduced receptor and/or postreceptor events, encompassed in the generic term "desensitization," may also contribute to the termination of hormonal responses. In the case of GCA-ANF receptors, the present results indicate that the rapid dissociation of membrane GCA-ANF receptor-ANF complexes at 37 "C may account at least in part for the termination of the cellular effects of ANF. The findings that there is a disproportionate increase in the off-rate of '"1I-ANF1-2e from GCA-ANF receptors a t physiological but not at subphysiological temperatures (Table 11) and that amiloride blocks this event (Fig. I) suggest that cellular metabolic events are involved in the regulation of the receptor-ligand dissociation. In membranes of bovine adrenal zona glomerulosa, amiloride has been shown to decrease the equilibrium dissociation constant ( kd) of GCA-ANF receptor at subphysiological temperatures, an effect that is opposite to and partially competitive with that of ATP (25,42). In addition, in the same preparation, ATP was shown to increase the kOff of 1ZsI-ANF1-2e from GC-ANF receptors to a greater extend that it increases the k,, (43). Finally, ATP and amiloride have been postulated to interact with the kinase domain of GCA-ANF receptors, which exerts a modulatory effect on guanylate cyclase activity and on receptor affinity (43,44). These data, as a whole, suggest that ATP is likely to have an important role in the very rapid dissociation of ANF from GCA-ANF receptors in intact cells. Although the contribution of other metabolic factors cannot be ruled out by the present data, a role for ATP in this process is supported by the finding that at subphysiological temperatures, in which there is limited production of ATP, the offrate of ligand from GCA-ANF receptors has a Qlo that is 2-3fold lower than that between 30 and 37 "C (Table 11).
To our knowledge, there have been no previous systematic studies on the temperature dependence of the dissociation of membrane receptor-ligand complexes in intact cells. In most instances such studies would be hampered by the internalization of these complexes, precluding a precise determination of the dissociation rate. It is noteworthy, however, that a very high off-rate of ligand a t 37 "C, similar in value to that described presently for GCA-ANF receptors, has been reported for polymeric IgA receptor-IgA complexes in Madin Darby canine kidney cells when internalization is blocked by hypertonic sucrose medium (30) and for noninternalizing insulin receptors in IM-9 lymphocytes (33). At 4 "C the dissociation rates of ANF from GCA-ANF and C-ANF receptors are similar and very slow, approximately l%/min or less (Table 11) (15, 45). However, at 37 "C there is a major disparity between the dissociation of ANF from C-ANF and GCA-ANF receptors. Results of previous studies have shown that at 37 "C the offset of ligand from surface C-ANF receptors in BAVSM and RC2 cells is < 0.05 min", the approximate internalization rate of C-ANF receptor-ligand complexes in these cells (15,45). In the present study, the slower component of the decay of radioligand from the surface of EGM cells (Fig. 4, panel A ) suggests that the off-rate from surface C-ANF receptors at 37 "C in these cells is also very slow. In contrast, the koff of ANF from GCA-ANF receptors in LGM cells at 37 "C is > 0.8 min" (Table 11). Unfortunately, because of the very high k,,,, the equilibrium dissociation constant ( k d ) or the onset rate ( L ) of ligand binding to GCA-ANF receptors at 37 "C cannot be experimentally determined. In all likelihood, however, the equilibrium dissociation constant at 37 "C is much higher for GCA-ANF than for C-ANF receptors. Consistent with this conclusion is the observation that at subphysiological tem-peratures, ATP increases the measured equilibrium dissociation constant ( kd) of radioligand binding to GC-ANF receptors in membranes from bovine adrenal zona glomerulosa (43). The present results point out that equilibrium or kinetic parameters of binding determined a t subphysiological temperatures and/or outside the natural environment of the cell do not necessarily reflect those in intact cells a t physiological temperatures.
The major differences in the dynamics of GCA-ANF and C-ANF receptors at physiological temperatures in intact cells are consistent with the specialized role of these receptors. The effectiveness of the removal of ANF from the circulation, a major role of C-ANF receptors, is determined by receptormediated internalization and lysosomal hydrolysis of ligand. This process is favored by a slow dissociation of surface receptor-ligand complexes. Indeed, a sluggish dissociation increases the residence time of the ligand on the receptor, allowing the relatively slow endocytosis of C-ANF receptors to deliver bound ANF to lysosomes, where it will be hydrolyzed (14, 15, 45). In spite of a very slow dissociation, the availability of unoccupied C-ANF receptors is assured under physiological conditions by the very high density of these receptors in vascular and renal tissues and the recycling of internalized C-ANF receptors to the membrane surface (12)(13)(14)(15)(16). However, mediation of the functional effects of ANF, the major role of GCA-ANF receptors, does not depend on internalization. Under this condition, a fast dissociation of ANF from membrane GCA-ANF receptors will contribute to hormonal stimulus-response homeostasis. When the free concentration of ANF falls, a faster dissociation favors a rapid termination of the response. The dissociated hormone can then be removed rapidly by the abundant C-ANF receptors.
Conversely, a faster dissociation increases the availability of unoccupied GCA-ANF receptors for ligand binding, favoring a heightened response when the free concentration of ANF raises. As discussed above, ATP is likely to play a major role in this homeostatic process because it increases ligand offset (and to a smaller extend onset) and magnifies the ANFinduced increase in guanylate cyclase activity. In combination, these events are likely to result in a higher efficacy of ANF effects when plasma concentration raises and a prompt termination of these effects when plasma concentration of ANF falls.