Antagonists of Bradykinin That Stabilize a G-protein-uncoupled State of the B2 Receptor Act as Inverse Agonists in Rat Myometrial Cells*

Several B2 bradykinin (BK) receptor-specific antago- nists including HOE140, NPC17731, and NPC567 exhibited negative intrinsic activity, which was observed as a decrease in basal phosphoinositide hydrolysis in primary cultures of rat myometrial cells, and this response was opposite to that elicited by the agonist BK. The order of potency of the antagonists in attenuating basal activity was essentially the same as that in competing both [‘HIBK and [‘H]NPC17731 for binding to B2 recep- tors on both intact rat myometrial cells and bovine myometrial membranes. We previously proposed a three- state model for the binding of agonists to G-protein-coupled B2 receptors in bovine myometrial membranes (Leeb-Lundberg, L. M. F. and Mathis, S. A. (1990) J. Biol. Chem. 265,9621-9627). This model was based on the ability of BK to promote the sequential formation of three receptor binding states where formation of the third, equilibrium state was blocked by Gpp(NH)p (guanyl-5”yl imidodiphosphate) identifying it as the G-protein- coupled state of the receptor. Here, we show that, in contrast to BK, these antagonists bound preferentially to a G-protein-uncoupled state of the receptor. These results indicate that B2 receptor antagonists that stabilize a G-protein-uncoupled state of the receptor act as inverse agonists. Furthermore, these results provide strong evidence that endogenous G-protein-coupled re- ceptors exhibit spontaneous activity in their natural en-vironment in the absence of agonist occupancy.

origin (3). PI hydrolysis through activation of phospholipase C represents one of the major intracellular signaling pathways that are mediated by these receptors in various cells (4)(5)(6)(7). This response appears to be regulated primarily by a G-protein (Y subunit of the aq family (8).
A number of B2 receptor peptide antagonists have been developed. A first generation of antagonists with moderate receptor affinities were developed around the crucial replacement of Pro7 with a D-aromatic residue (9, 10). Later, a second generation of antagonists were developed specifically to incorporate a conformationally restricted / 3 t u r n in the C-terminal residues of the BK analogs, which resulted in considerably higher affinities for the B2 receptor (11)(12)(13)(14)(15). Classical drug theories state that agonists either fully or partially stimulate the receptor-mediated functional response, while antagonists block the interaction of the agonists with the receptor but lack any intrinsic activity themselves. Subsequent theories propose that unoccupied receptors can spontaneously isomerize between "inactive" and "activated" states and that antagonistic drugs either do not distinguish between these states (neutral antagonists)  Here, we show that several bradykinin antagonists behave as inverse agonists on PI hydrolysis in primary cultures of rat myometrial cells. This activity was correlated with the ability of these antagonists to stabilize a G-protein-uncoupled state of this receptor in bovine myometrial membranes. These results reinforce our previously proposed three-state model for the binding of agonists to the G-protein-coupled B2 receptor (2). EXPERIMENTAL PROCEDURES Cell and Membrane Preparations-Primary cultures of rat myometrial cells and bovine myometrial membranes were prepared as described by Tropea et al. (7) and Leeb-Lundberg and Mathis (2), respectively. Protein concentration was determined by the method of Bradford (31) using bovine serum albumin as a standard.
Receptor Binding-Radioligand binding assays with the agonist [3H]BK and the antagonist [3HlNPC17731 were performed essentially as described (2, 7). In short, intact cell binding was assayed in DMEM, 0.1% bovine serum albumin including the protease inhibitors bacitracin (140 pg/ml) and 1,lO-phenanthroline (1 mM) at 4 "C for 90 min. Membrane binding was assayed in 25 m M TES, pH 6.8,0.5 m M EDTA, 1 m M MgCl, including the above inhibitors, at 25 "C for 60 min. In both protocols, nonspecific binding was determined in the presence of 1 p~ BK.
In some experiments, any receptor-bound BK was removed prior to binding assays on intact cells by incubating cells with a low pH buffer (50 m M glycine-HC1, pH 3.0) for 3 min at 4 "C followed by two l-ml washes with phosphate-buffered saline containing 0.3% bovine serum albumin as described by  and Munoz et al. (33).
PI Hydrolysis-Cells were assayed essentially as described (7), with a few modifications. Briefly, confluent cells were incubated with 10 pCUm1 my~-[~H]inositol in DMEM, 1% heat-inactivated fetal bovine serum at 37 "C for 24 h in 10% CO,. Prior to experimentation, the cells were washed four times with 1 ml of DMEM and incubated in DMEM, 50 m M LiCl for 30 min. Following replacement with 2 ml of the same medium, the cells were incubated at 37 "C with various agonists and antagonists for 20 min. Inositol phosphates were then extracted and isolated using anion exchange chromatography.
Analysis of Cell-derived BK-Medium (40 ml) incubated as described above for PI hydrolysis and low pH buffer wash of incubated cells were supplemented with -lo4 dpm (-58 pg) r3H1BK (internal standard). The supplemented preparations were then processed on a C,, SepPak cartridge (Waters Associates). The recovery of r3H1BK was 290%. The BK concentration was then assayed by RIA according to a procedure described by the supplier (Peninsula, Belmont, CA). Identification of the internal standard in the RIA confirmed absence of BK degradation during preparation.

RESULTS AND DISCUSSION
Primary cultures of rat myometrial cells contain a large number of a single type of B2 receptor as determined by the high affinity specific binding of the agonist I3H1BK (185,000 2 47,000 siteskell, KD = 1.8 2 0.2 m; n = 3) (7). Fig. lA shows that this binding was specifically competed by the antagonists HOE140, NPC17731, and NPC567 with an order of potency typical for a B2 receptor. The antagonist [3HlNPC17731 identified the same B2 receptor as determined by the amount (230,000 2 5,000 siteslcell; n = 3) and the affinity (K, = 0.9 2 0.2 m) of binding as well as by the specificity of agonist and antagonist competition (Fig. 1B).
Addition of BK to rat myometrial cells resulted in a dramatic and dose-dependent increase in PI hydrolysis indicating that BK acts as a full agonist on these cells (7) (Fig. 2, upper panel). This response was clearly mediated through the B2 receptor as it was inhibited by the above antagonists with an appropriate Interestingly, each of these antagonists exhibited negative intrinsic activity on their own (in the absence of agonist) by significantly reducing the basal level of PI hydrolysis (Fig. 2,  lower panel). As expected for receptor-mediated events, these reductions were dose-dependent and saturable. Furthermore, the maximal reduction (3540%) in the basal PI hydrolysis was virtually the same for all the antagonists, and the potency of each in reducing the basal activity was similar to their potency in competing for both L3H1BK and [3HlNPC17731 binding to these cells. By exhibiting negative intrinsic activity, these antagonists appear to behave as inverse agonists such as have been observed with antagonists in some ion channel receptor systems (18).
An alternative explanation for the decrease in basal activity is simple antagonism of stimulation by cell-derived BK released during the assay. As shown by the dose-response curve in Fig. 2, a minimum of 0.1 n~ BK was required to elevate PI hydrolysis in these cells. By RIA, which is sensitive to as little as 0.25 PM, no BK was detected in medium previously incubated with cells as described for the assay of PI hydrolysis (conditioned medium). The same result was obtained from RIA of collected low pH buffer used to strip any bound cell-derived BK from the extracellular surface of incubated cells. Furthermore, conditioned medium did not inhibit either L3H]BK or L3H1NPC17731 binding to naive cells. In addition, we found no increase in the binding of either radioligand to incubated cells that had been stripped with the low pH buffer (data not shown). Thus, the amount, if any, of cell-derived BK in the medium during the functional assay is <0.25% of that required for stimulation of PI hydrolysis. Consequently, the negative intrinsic activity of these antagonists cannot be explained by inhibition of the action of cell-derived BK.
Inverse agonism requires that the unoccupied B2 receptor spontaneously isomerizes between an inactive conformational state, which can be stabilized by antagonists, and an activated state, stabilized by agonists. Previously, we proposed a threestate model for BK binding to GTP-sensitive B2 receptors in bovine myometrial membranes (Fig. 3) (2). According to this model, BK (L) and receptor (R) initially form a complex from which BK dissociates rapidly, which is termed LR. In a timedependent manner, the binding complex becomes stabilized through the conversion of LR through LR* to a state from which BK dissociates very slowly, termed LR**. Inclusion of Gpp-(NH)p, a nonhydrolyzable analog of GTP, completely blocks the formation of LR**. We believe LR" is the state in which the receptor exists while functionally coupled with the G-protein, hence it is termed LR**G. BK dissociates more slowly from R**G than from either R or R'. Thus, just as agonist receptor binding promotes increased G-protein receptor coupling, G-protein coupling apparently results in increased agonist affinity. Unlike the ternary complex model described by DeLean et al. (34) to explain the interaction of agonists and antagonists with Gprotein-coupled P-adrenergic receptors, our model contains the additional receptor state R' to which the agonist can be bound.
Considering the relative position of R* in the binding sequence, we originally suggested that this state represents a "precoupled" or activated form of the receptor not yet functionally coupled to the G-protein (2). Conceptually, our model derived from studies in bovine myometrial membranes can accommodate the inverse agonism observed in rat myometrial cells by equating R with the antagonist-stabilized inactive receptor state and R' with the agonist-stabilized activated receptor state.
We compared the pharmacological characteristics of agonist and antagonist binding in these two receptor systems. As in rat myometrial cells, [3HlBK and L3H]NPC17731 identified a n equal number of binding sites in the bovine membranes.' Fig. 4 (A-G) shows that the order of potency of BK, HOE140, NPC17731, and NPC567 in competing for L3H1BK binding was virtually the same as competing for [3HlNPC17731 binding, indicating that these two radioligands identify the same receptor in these membranes. Furthermore, the specificity of agonist and antagonist binding in rat myometrial cells and bovine myometrial membranes was also very similar (compare Figs. 1 and  4), indicating that these two systems contain B2 receptors that are at least pharmacologically virtually identical.
Since the agonists bind with higher affinity to the G-proteincoupled state of the receptor, in order to exhibit negative intrinsic activity inverse agonists must bind preferentially to a G-protein-uncoupled state. To demonstrate this experimentally, we compared agonist and antagonist binding in bovine membranes in the absence and presence of Gpp(NH)p. Inclusion of 10 p~ Gpp(NH)p resulted in a rightward shift (6.3-fold) in the competition curve for BK when competing for [3H]NPC17731 binding (Fig. 4A) when competing [3HlBK binding (Fig. 40). This observation is not surprising since BK binds in a manner identical to that of F3H1BK and, consequently, is subject to the same heterogeneity in the binding affinity to different conformational states of the receptor. The same reasoning may be used in interpreting the absence of a Gpp(NH)p-induced shift in the competition curves for the antagonists HOE140 (Fig. 4B) and NPC17731 (Fig. 4C) when competing [3HlNPC17731 binding. In contrast, when competing for C3H1BK binding, Gpp(NH)p induced a leftward shift in the competition curves for HOE140 (3.8-fold; Fig. a), NPC17731 (5.2-fold;Fig. @), and NPC567 (1.5-fold; Fig. 4G). These results indicate that agonists and antagonists prefer to bind to different states of the receptor. The effect of Gpp(NH)p shows that the antagonists stabilize a G-protein-uncoupled state.
This report describes several important observations that directly relate to the nature of the different agonist binding states of the B2 receptor (2). Based on our observation of B2 receptor antagonists with negative intrinsic activity, we conclude that, as in some other receptor systems (181, this receptor can spontaneously isomerize between an inactive and an activated conformational state in the absence of agonist binding and that when the receptor is in the activated state, it can couple with a G-protein to trigger a functional response. Results on which our model was originally based indicate that only after agonist binding for very short time periods (seconds) does R represent the major receptor state (2). On the other hand, at equilibrium agonist binding, in the presence of Gpp(NH)p, which prevents R* from isomerizing into R**G, R* represents over 50% of the total agonist binding. Thus, over time agonist binding clearly favors formation of R" at the expense of R. In the absence of Gpp(NH)p, the agonist rapidly converts R* to R**G, the G-protein-coupled state of the receptor. Consequently, we conclude that R represents the inactive state of the receptor and R* represents the agonist-stabilized activated receptor state, which is not yet functionally coupled to the Gprotein.  (-1 m) in the absence (0) and presence (0) of 10 Gpp(NH)p. Dissociation was initiated by 5-fold dilution and making the suspension 1 prd BK, and assays were terminated at various times as described under "Experimental Procedures." Other conditions were as described in Fig. 4.
The results shown are the averages 2 S.E. of three experiments with each point performed in duplicate.
As predicted by the model for a ligand that stabilizes the inactive R state, addition of Gpp(NH)p to block the agonistpromoted formation of R**G and shift the equilibrium to the inactive R state increased the affinity of the antagonists for the receptor (Fig. 4). This antagonist selectivity is further supported by the fact that addition of Gpp(NH)p in the absence of any agonist had no significant effect on the dissociation rate of [3H]NPC17731 binding (Fig. 5). Even though these antagonists inhibit BK binding, cross-linking studies indicate that the binding sites for NPC17731 and HOE140 in the B2 receptor are not identical to that for BK, providing one possible explanation as to why these two classes of ligands stabilize different conformational states of the receptor?
In order to accommodate the observations of constitutive activity for some G-protein-coupled receptors in membranes and reconstituted systems (20-24) and in transfected systems using either wild-type (25,261 (2) and elaborated upon in this report. The resemblance between these two models suggests that G-protein-coupled receptors behave similarly regardless of whether they accept peptidergic or non-peptidergic ligands.
In general, second generation B2 receptor antagonists, of which HOE140 and NPC17731 are representative examples, are at least 10-fold more potent in antagonizing BK-stimulated functional responses than first generation antagonists, of which NPC567 is a representative example (36). This difference in antagonistic potency was not directly reflected in the potency of these antagonists to compete for [3H]BK binding on the myometrial preparations where NPC567 was about equipotent with HOE140 and NPC17731. In contrast, NPC567 was approximately 16-and 13-fold less potent than HOE140 and NPC17731, respectively, in inhibiting basal PI hydrolysis. Furthermore, while the Gpp(NH)p-induced shift of the NPC567 competition curve was 1.5-fold, the HOE140 and NPC17731 curves were shifted 3.8-and 5.2-fold, respectively. Thus, the ability of antagonists to stabilize the inactive conformation of the B2 receptor is a more accurate parameter of antagonistic activity in this system than their ability to displace BK binding to the receptor.
In summary, the results reported here serve to further define the nature of each of the binding states in our model. This was possible due to 1) B2 receptor ligands exhibiting negative intrinsic activity, 2) identification of a cell system with a SUE-ciently high number of B2 receptors that exist spontaneously in a n activated state, and 3) the relatively high afiinity binding of BK to both G-protein-coupled and -uncoupled states of the B2 receptor. The few previous studies with other G-proteincoupled receptors in intact cells have all used cells transfected with either wild-type or constitutively activated mutant receptors. To our knowledge, this is the first study demonstrating in a primary cultured cell system containing an endogenous Gprotein-coupled receptor that inverse agonism in this superfamily of receptors is not an artifact but a natural feature of the receptor system, and this receptor behavior is supported by our proposed model.