Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The super family of G-protein-coupled receptors (GPCRs) share a common structural property of a single polypeptide chain traversing the membrane bilayer using seven -helical domains. These seven helices form a hydrophilic ligand binding pocket to which the endogenous agonists are thought to interact with the receptor protein. Adrenergic receptors (ARs) comprise a subfamily of related receptors that mediate the effects of the sympathetic nervous system by binding the endogenous agonists, epinephrine and norepinephrine (Hwa et al., 1996a). ARs are subdivided into three classes and each class has three known receptor subtypes (₁, ₂, ₃; _2a, _2b, _2c; _1a, _1b, _1d). Activation of these GPCRs is thought to involve agonist-mediated changes in the receptor tertiary structure. For example, during the photoactivation process, rigid body movements of transmembrane domains (TMDs) three and six have been described for the prototypical GPCR, rhodopsin (Farahbakhsh et al., 1995; Altenbach et al., 1996). In addition, investigations of a related seven TMD receptor, bacteriorhodopsin, have observed physical changes in TMDs six and seven (Subramaniam et al., 1993), although this receptor is not G-protein-coupled, but is involved with light-dependent bacterial proton pumping. The activational mechanism of rhodopsin involves the disruption of a salt-bridge constraint between a TMD three E113 and a K296 in TMD seven, which forms a protonated Schiff base with retinal (Robinson et al., 1992). Light-induced isomerization of cis-retinal to the all trans form breaks the rhodopsin salt-bridge leading to receptor activation. However, for any member of the GPCR superfamily other than rhodopsin, the molecular mechanism of receptor activation is not known.

We have demonstrated previously that activation of ₁-ARs is conserved to the rhodopsin paradigm, via disruption of a constraining salt-bridge between a conserved aspartic acid in TMD three (D125) and a lysine residue (K331) in TMD seven (Porter et al., 1996). Using site-directed mutagenesis, elimination of the charged amino acids that comprise this salt-bridge causes the _1b-AR to become constitutively active, mimicking the activated state of the wild-type (wt) receptor. From this, we hypothesize the mechanistic process of agonist-dependent receptor activation involves a competition for the negatively charged D125 by the protonated amine of the endogenous catecholamine agonist and the positively charged K331. Thus, competition by the basic amine of the AR agonist disrupts the constraining _1b-AR salt-bridge and initiates the activation process of the receptor (Porter et al., 1998). However, there is no direct evidence that K331 is interacting with D125, only that their "charged state" is important in receptor function.

The pH level at which any ionizable groups in an aqueous solution are half-ionized is defined as the negative logarithm of the acid dissociation constant (pK_a). For complex membrane proteins such as GPCRs, the pK_a of amino acids in the receptor depend on the microenvironment of these ionizable groups, known as the field effect in protein chemistry. The ₁-AR salt-bridge is an ionizable pair of amino acids in the hydrophilic binding pocket and each counterion would stabilize its partner's charged state, affecting their acid strength or pK_a. Eliminating the charge at one position of the salt-bridge, thereby leaving its paired amino acid unbuffered should change the apparent counterion's pK_a value. To test this hypothesis and support K331 interaction as an ionic constraint with D125 in a functionally relevant manner, the acid strength of D125 was quantitated for wt and K331 mutants of the _1b-AR. Results from these studies substantiate the hypothesis of an _1b-AR salt-bridge and its role in the receptor activation process. In addition, pK_a differences from the wt receptor were used to calculate the free energy needed to disrupt the _1b-AR salt-bridge. The estimated strength calculated for the _1b-AR salt-bridge (1 kcal/mol) is consistent with a mechanism whereby the free energy of agonist binding to the receptor causes disruption of this ionic constraint.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Site-Directed Mutagenesis. Site-directed mutagenesis was performed on a M13 mp19 hamster _1b-AR construct using the oligonucleotide-mediated double primer method (Sambrook et al., 1989) and as described previously (Porter et al., 1996). DNA was purified and sequenced by the dideoxy method to verify the mutation. Mutated _1b-AR inserts were removed from the phage M13 mp19 vector then subcloned into a eucaryotic expression vector, pMT2' as described previously (Porter et al., 1996). The full-length plasmid DNA was again sequenced to verify the mutation.

Cell Culture and Transfection. COS-1 cells (ATCC) were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, and the transient transfection was performed as described previously using the DEAE-dextran method (Porter et al., 1996). Cells are used 60 h post-transfection. The poorly expressed D125A _1b-AR mutant can be up-regulated in receptor density by adding 1 µM prazosin to the medium 24 h before harvesting.

Quantification of Phosphotidyl Inositol Hydrolysis. Basal measurements for phospholipase C-dependent inositol 1,4,5-triphosphate (IP₃) production was determined by a [³H]IP₃ radioreceptor assay (DuPont) using the same preassay condition as described previously (Porter et al., 1996). Sixty-millimeter dishes containing 10⁶ cells were transiently transfected. Saturation binding studies were conducted on membranes prepared from parallel dishes using the same DNA transfection cocktail to determine the amount of receptor expression.

Radioligand Binding. Membranes from transfected COS-1 cells were prepared as described previously (Perez et al., 1991). Pharmacological profiles of expressed _1b-ARs were determined by saturation and/or competition binding experiments using the selective ₁-AR antagonist (±)--([¹²⁵I]iodo-4-hydroxyphenyl)-ethyl-aminomethyl-tetralone ([¹²⁵I]HEAT) as the radiolabel. All binding experiments were performed as described previously (Porter et al., 1996). Binding curves were generated using iterative nonlinear regression analysis (Motulsky and Ransnas, 1987). Protein concentrations were measured using the method of Bradford (Bradford, 1976). Buffers used were HEM (20 mM HEPES, 12.5 mM MgCl₂, 1.5 mM EGTA) over the pH range 7 to 8.5, whereas MEM (20 mM MES, 12.5 mM MgCl₂, 1.5 mM EGTA) was used in the pH range 5 to 6.5. The pH of binding assays was remeasured after addition of membranes and drugs to ensure correct pH assessment.

Apparent pK_a and Free-Energy Calculations. Estimating the apparent acid dissociation constant (K_a) for D125 of the _1b-AR was modified from previously described studies for other biogenic amine receptors (Williamson and Strange, 1990). Briefly, a single ionizable group (D125) located in TMD three of the _1b-AR, is known to be important for the binding of protonated amine AR agonists such as epinephrine (Porter et al., 1996). Calculating the ligand dissociation equilibrium binding constant (K_i) from competition binding experiments can quantitate this affinity of the ligand for the receptor (eq. 1)

(1)

where R stands for the receptor, L for the ligand, and K_i for the dissociation equilibrium binding constant represented by the ratio of k₁/k₊₁.

(2)

(3)

1

1

(4)

(5)

(6)

Statistical Analysis. For each individual experiment, the fitted iterative nonlinear regression curve that best represented the data was determined using a partial f-test, F = [(SS1-SS2)/SS2]/[(DF1-DF2)/DF2], where SS = sum of the squares and DF = degrees of freedom (P < .05). A runs test determined if the data adhered to the modeled equation (P > .05). Significance between groups was tested using an unpaired two-tailed Student's t test (P < .05). All values are reported as the mean ± S.E. for n experiments, each performed in duplicate.

Materials. ()-Epinephrine, Sigma; [³H]IP₃ kit, [¹²⁵I]HEAT, DuPont-New England Nuclear.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

_1b-AR Epinephrine Competition Binding and Assignment to K_iH and K_iL. Epinephrine competition curves for the wt and K331Q _1b-AR mutant are shown in Fig. 1. Both competition curves statistically fit best to a one-site binding model. The higher affinity seen in the K331Q _1b-AR mutant is intrinsic to the receptor as assessed by adding 0.1 mM GppNHp to the binding assay (data not shown). The one-site model fitted for the wt or K331Q ₁-AR is attributable to its transient expression in COS-1 cells, nonlimiting G-protein concentrations, and its specific coupling to G_q, which has a low intrinsic GTPase activity (Ross, 1995). A one-site fit is common in ₁-ARs and reported for all previously described constitutively active single mutations of the _1b-AR (Kjelsberg et al., 1992; Hwa et al., 1996b; Perez et al., 1996; Scheer et al., 1997). When we increased constitutive activity by combining these single mutations in the same receptor, we then could achieve two-site binding curves (Hwa et al., 1997). Thus, K_iH is assigned to the K331Q mutant receptor and K_iL to the wt _1b-AR. Changes in both J and J constants should be inherent in the actual K_i values experimentally measured at different pH values. Although we realize that the K331Q mutant is not fully activated and conversely, that the wt receptor is not fully inactive, it is assumed that each receptor represents a majority of either the K_iH or K_iL states of the _1b-AR given the one-site fit.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Epinephrine competition curves for wt and K331Q mutant 1b-adrenergic receptors. Epinephrine competition of specific [125I]HEAT binding to wt () or K331Q () 1b-ARs at pH 7.5. Nonlinear regression curves fit best to a one-site model (wt: DF1 = 57, DF2 = 55, SS1 = 4351, SS2 = 4217, F = 0.87, P < .05; K331Q: DF1 = 53, DF2 = 51, SS1 = 1820, SS2 = 1699, F = 1.82, P < .05). The binding affinity (Ki) of epinephrine (0.37 ± 0.01 µM) for the K331Q mutant 1b-AR was significantly greater (P < .05) than for the wt receptor (1.8 ± 0.1 µM). Measurements are presented as the mean ± S.E. for n = 3-4 experiments performed in duplicate.

Constitutive Activity of the _1b K331 AR Mutants. The increased epinephrine affinity calculated for the K331Q _1b-AR mutant and assignment to K_iH is based on whether the receptor is indeed constitutively active representing the R* state. We have previously shown that K331A and K331E _1b-AR mutants are constitutively active displaying both high agonist affinities and elevated basal IP₃ levels (Porter et al., 1996). In current studies, the K331D _1b-AR also constitutively activates IP₃ production (Porter and Perez, 1999). These two parameters of constitutive activity, high receptor agonist affinity and elevated basal second messenger levels, are predicted from the revised ternary complex model of GPCR activation (Samama et al., 1993). Theoretically, all mutations at K331 of the _1b-AR that eliminate the endogenous positive charge and therefore disrupt the ionic bond with D125 should produce a constitutive phenotype. To confirm this, basal IP₃ production was measured for both K331Q and K331H receptor mutants at physiological pH (7.3) and compared with the wt _1b-AR (Fig. 2). Studies were performed at equivalent receptor expression. There was a significant increase (P < .05) in the amount of IP₃ produced for the K331Q receptor mutation (29.3 ± 2.5 pmol/fmol receptor) when compared with the wt _1b-AR (15.8 ± 1.0 pmol/fmol receptor), consistent with a mechanism involving disruption of an ionic constraint. However, there was no significant increase from wt receptor for IP₃ production by the K331H _1b-AR mutant (17.8 ± 2.1 pmol/fmol receptor).

View larger version (35K): [in this window] [in a new window]   Fig. 2.   Agonist-independent production of IP3 for K331 mutations of the 1b-AR. The amount of basal IP3 generated from COS-1 cells (106 cells) transfected with either the wt, K331H, or K331Q 1b-ARs was quantified and normalized to receptor number determined from parallel saturation binding experiments. Assays were done in DMEM medium with no serum at pH 7.3. There was a significant increase (P < .05) in the basal amount of IP3 generated per receptor for the K331Q (29.3 ± 2.5 pmol/fmol) mutant receptor when compared with either K331H (17.8 ± 2.1 pmol/fmol) or the wt (15.8 ± 1.0 pmol/fmol) 1b-AR. Receptor density for the wt 1b-AR was titrated down by using less DNA in the transfection mixture to approximate the expression level of the mutant receptors. Actual receptor density was 1.7 ± 0.1 pmol/mg protein for the wt 1b-AR, 3.3 ± 1.7 pmol/mg protein for the K331H mutant receptor and 1.5 ± 0.8 pmol/mg protein for the K331Q receptor mutant. Data points are presented as the mean ± S.E. for n = 3 transfections of 10 replicates each.

_1b-AR Antagonist Binding. Saturation binding analysis was also performed at different pH values to ascertain any changes that may occur in the dissociation equilibrium binding constant (K_D) of the radiolabel, which is used to calculate the epinephrine binding affinity for the receptor. The specific ₁-AR antagonist, [¹²⁵I]HEAT, was used to radiolabel _1b-ARs transiently expressed on COS-1 cell membranes. Saturation binding curves were significant for a one-site binding model with the number of receptors unaffected by alterations in pH (data not shown). The affinity of the radiolabel was also not significantly affected by pH for either the wt (Fig. 3) or K331Q mutant _1b-AR (data not shown). This suggests that the ionized state of D125 or other acidic residues involved in agonist binding are not crucial for the affinity of the receptor antagonist. These results are consistent with previously published work where the affinity of AR antagonists for D125A or D125K _1b-AR mutants also did not significantly change (Porter et al., 1996). However, affinity of the ₁-AR agonist, epinephrine, was sensitive to changes in pH, suggesting that the ionized state of D125 and other acidic residues does indeed influence the binding of this ligand (Fig. 3).

View larger version (23K): [in this window] [in a new window]   Fig. 3.   pH-Dependence of [125I]HEAT and epinephrine binding affinity for 1b-adrenergic receptors. Affinity values of epinephrine () and the radiolabled 1-AR antagonist, [125I]HEAT () for the wt 1b-AR was calculated from competition and/or saturation binding experiments performed at increasing pH. Epinephrine affinity values for the 1b-D125A-AR mutant () are shown to demonstrate the influence of D125 on the acidic pKa values. Calculated pKa values (arrows) for the D125A receptor mutant (5.98 ± 0.04) was significantly different (P < .05) from the wt 1b-AR (6.25 ± 0.02). Measurements are presented as the mean ± S.E. for n = 2-8 experiments performed in duplicate.

Calculation and Assignment of Acidic pK_a. To further demonstrate the acidic pK_a of the wt _1b-AR receptor is influenced by D125 and that changes in acidic pK_a can be assigned to D125, the mutant _1b-D125A-AR was used in epinephrine competition binding studies. As shown in Fig. 3, this receptor mutant displayed a decreased epinephrine binding affinity, a less sensitive pH profile, and a lower acidic pK_a value than the wt receptor. Because the pK_a value of D125 in the wt receptor is approximately 6.25 ± 0.02 (Fig. 4; Table 1), elimination of this weaker acidic residue in the D125A mutant receptor would shift the resulting acidic pH profile to the left.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   pH-dependent epinephrine affinity values for wt, K331Q, and K331H mutant 1b-adrenergic receptors. Panel A, epinephrine affinity values are calculated from competition binding experiments performed at different pH values for the wt () and K331Q () mutant 1b-AR. The pKa (arrows) calculated for the K331Q receptor mutant (7.07 ± 0.05) was significantly different (P < .05) from that of the wt receptor (6.25 ± 0.02). Panel B, pH-dependent epinephrine affinity values for wt () and K331H () mutant 1b-ARs. The calculated pKa (arrow) for the K331H 1b-AR mutant (6.45 ± 0.03) was not significantly different (P > .05) from the wt receptor. Measurements are presented as the mean ± S.E. for n = 5-8 experiments performed in duplicate.

View this table: [in this window] [in a new window]   TABLE 1 Acid dissociation constants and changes in free energy for 1b-adrenergic receptors Estimation of the acid dissociation constant (Ka) for the wt and salt-bridge mutants of the 1b-AR were calculated from epinephrine affinity values using the equation, Ki(obs) = Ki(1 + [H+]/Ka) described in the experimental procedures. Ka changes of the K331 mutations from wt receptor were used to calculate the free energy of the constraining 1b-AR salt bridge using the equation G = RT × ln[Ka(wt)Ka(mutant)] described in Materials and Methods.

View larger version (26K): [in this window] [in a new window]   Fig. 5.   pH-dependent epinephrine affinity values for wt and K331 1b-adrenergic receptor mutants. Epinephrine affinity values are calculated from competition binding experiments performed at different pH values for the wt (), K331A (), K331M (), K331L (), K331D (), K331W (), and K331E (×) 1b-AR mutants. The pKa (arrows) calculated for each mutant receptor is listed in Table 1. The binding affinities (Ki) of epinephrine for the K331 mutant 1b-ARs was significantly greater (P < .05) than for the wt receptor. Measurements are presented as the mean ± S.E. for n = 2-8 experiments performed in duplicate.

View larger version (25K): [in this window] [in a new window]   Fig. 6.   pH-dependent epinephrine affinity values for wt, K331D, D125A, and the D125K/K331D switch mutant. Epinephrine affinity values are calculated from competition binding experiments performed at different pH values for the wt (), K331D (), D125A (), and D125K/K331D switch (×) 1b-AR mutants. The pKa (arrows) calculated for each mutant receptor is listed in Table 1. The binding affinities (Ki) of epinephrine as well as the calculated pKa for the K331D single mutant was significantly greater (P < .05) than for the wt receptor. However, the binding affinities of epinephrine and calculated pKa of the switch mutant is not significantly different from the single D125A receptor mutant. Measurements are presented as the mean ± S.E. for n = 2-8 experiments performed in duplicate.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The positive charge on biogenic amine ligands is the most distinguishing moiety of this class of receptor agonists. This cationic site suggests the existence of a complementary anionic site on the receptor important perhaps for both binding and high efficacy of agonism. Indeed, mutation of an aspartic acid in TMD three of the ₂-AR (D113) (Strader et al., 1988), the _2a-AR (D113) (Wang et al., 1991), or the _1b-AR (D125) (Porter et al., 1996), have all resulted in lower or nondetectable binding affinity of the protonated amine receptor agonists for these mutant ARs with corresponding changes in the ligands ability to invoke agonism.

We have hypothesized previously that D125 in TM3 of the _1b-AR is directly involved mechanistically in the activation process. Specifically, we proposed that a salt-bridge constraint is formed between K331 in TMD seven and D125 in TMD three of the _1b-AR and that this ionic bond is required to maintain the inactive state of the receptor (Porter et al., 1996). When an agonist binds, a competition is established between K331 and the protonated amine of the ligand for neutrality with D125, resulting in salt-bridge breakage. Although salt-bridge disruption does not lead to full receptor activation as assessed by the partial constitutive behavior of the salt-bridge mutants (Porter et al., 1996) or the poor ability of basic salts to act as ₁-AR agonists, we postulate that it is an initial step of the activation process, because basic salts can potentiate the agonism of partial agonists (Porter et al., 1998). This mechanism may also be linked through hydrogen bonding networks to the proposed protonation/deprotonation of an aspartic acid residue in the DRY motif (Scheer et al., 1997). This paradigm is conserved at least partially to rhodopsin where the ₁-ARs are the closest phylogenetic neighbor, based on sequence alignment, outside of the opsin family. As additional proof of this mechanism, the postulated close association of K331 with D125 of the _1b-AR should effect each other's pK_a. Predictive models for protonation equilibria in proteins relies primarily on the concept that pK_a shifts result from electrostatic interactions (Antosiewicz and McCammon, 1996). Therefore, if the _1b-AR salt-bridge formation is valid, mutation of K331 to residues that eliminate the positive charge should change the pK_a of D125, making it a weaker acid. Because the anionic site of the _1b-AR is dependent on the ionization of D125, then binding of the protonated amine agonist should be effected by pH changes in the pK_a range of D125. This forms the basis for eqs. 1 and 2 of our experimental system.

An assumption in eq. 1 is that no significant receptor agonist binding occurs when D125 is fully protonated. This assumption would have bearing for an absolute calculation of the D125 pK_a. However, many reports have confirmed that significant docking still occurs when the equivalent aspartic acid is mutated (Strader et al., 1988; Wang et al., 1991; Porter et al., 1996). This is apparent in Fig. 3 where the D125A _1b-AR mutant has lower, but still significant, binding for epinephrine at all pH values. Furthermore, other acidic residues not implicated as direct binding contacts but critically involved in the correct folding of the receptor tertiary structure and, thus, formation of the binding pocket will contribute to an acidic pH binding profile. Because we are not comparing absolute numbers but differences in the pK_a between wt and _1b-AR mutants, this assumption does affect our results. Equation 1 is also simplified by negating any contribution from the ionized state of the receptor agonist. Again, because we are measuring differences between wt and mutant receptors, ionization of the ligand would be a constant and maximal given a pK_a of 9 to 10 for secondary amines.

To demonstrate that eqs. 3 and 4 used in data modeling are valid, and the wt and constitutively active receptor mutant's affinity can be used in independent equations to calculate pK_a1 and pK_a2, eq. 5 was derived from eq. 3 to account for how changes in the isomerization constants can influence the pH binding profile of the wt _1b-AR. Figure 7 shows theoretical curves at different J/J ratios generated using the experimental wt _1b-AR data. Increasing the J/J ratio to 2.5 results in a theoretical curve that fits the experimental data generated from the K331Q receptor mutant (r² = 0.95), suggesting that eqs. 3 and 4 are valid. Therefore, changes in J/J not only account for epinephrine's higher binding affinity, but also the increased pK_a2 value of the K331Q _1b-AR mutant. Therefore, the K331Q receptors pH profile in Fig. 4 shifts both upward (because of constitutive activity) and to the right (because of decreased acid strength).

View larger version (22K): [in this window] [in a new window]   Fig. 7.   The effect of changes in J/J on Ki(obs) versus pH. Equation 5 was used to generate theoretical curves with J/J ratios of 0.5, 1, 2.5, and 6 using the data obtained for the wt 1b-AR () with a Ki value of 1.99 × 106 (antilog of 5.7) and Ka1 of 6.31 × 107 (antilog of 6.2). A J/J ratio of 2.5 was found to best fit the data for the K331Q mutant 1b-AR () with a 0.95 correlation coefficient. An elevated J/J ratio increases both agonist affinity and pKa1 values (arrows).

Experimental measurement of the pK_a for D125 in the wt receptor was 6.25 ± 0.02. This is in a comparable range (5.2-6.1) for other reported aspartic acid groups in TMD three of biogenic amine receptors (Ehlert and Delen, 1990; Williamson and Strange, 1990) and for light-activated rhodopsin (5.9), which contains E113 in TMD three (Cohen et al., 1992). Mutation of K331 that resulted in the disruption of the positive charge resulted in significant elevation of D125 pK_a from the wt receptor, consistent with the original hypothesis that a salt-bridge is formed between K331 and D125. Unexpectedly, substitution of K331 with a histidine did not change the acidic profile from that of the wt receptor. This suggests that the K331H _1b-AR has maintained an ionic bond with D125. Although we initially made this mutation to see if we could titrate the histidine positive charge and see a conversion to an activated receptor with increasing pH, these results support our previous salt-bridge receptor mutant characterization that suggests disruption of this ionic constraint is a discriminating step in the activation mechanism of _1b-ARs (Porter et al., 1996). In addition, constitutive activation of the K331Q receptor mutant suggests no hydrogen bond component of the _1b-AR salt-bridge and implies that this constraint is completely ionic in nature. Although the functional group of free histidine has a pK_a of 6.0 in water, when incorporated into a protein, this value may substantially change to a pK_a typically around 7.0 (Barker, 1971). This is especially true if histidine is participating in an ionic bond where its pK_a would be elevated, thereby increasing its base strength. The results suggest that in the context of the _1b-AR binding pocket, the substituted histidine at position 331 was protonated over the pH range tested and can substitute for K331 in maintaining an ionic bond with D125 consistent with its nonconstitutive nature of IP₃ release (Fig. 2). In addition, the pK_a of K331H could have been raised above 8.5, because no conversion to the high affinity binding site was observed for the _1b-K331H-AR in the pH range of 5 to 8.5. This is comparable with results obtained using a K296H mutant of the rhodopsin receptor where histidine also was able to substitute for lysine in the rhodopsin salt-bridge and was not constitutively active from pH 5 to 9.5 (Cohen et al., 1993).

In the current study, we assign changes in the acidic pK_a to D125 based on changes in the acid dissociation constant for the _1b-D125A-AR, the direct role D125 has in epinephrine binding to the _1b-AR (Porter et al., 1996), and the basic shift in the acidic profile of K331 _1b-AR mutants. There are no other known acidic residues in the _1b-AR binding pocket directly involved with agonist binding. However, there are two other important aspartic acid residues that may be protonated/deprotonated during the activation process of the _1b-AR. D81 in TMD two of the _1b-AR, a highly conserved residue in all GPCRs, is buried two-thirds down in the hydrophobic transmembrane environment and is likely not accessible or effected by changes in pH introduced by water. This is substantiated by previous work where the analogous mutation in the rhodopsin receptor, D83, displayed a wt pH profile for activation (Cohen et al., 1993). Another highly conserved acidic residue, E134 in the ERY motif at the cytosolic end of TMD three of the rhodopsin receptor, has been shown to become protonated during the receptor activation process (Arnis et al., 1994). This same mechanism is hypothesized to occur for the analogous acidic residue (D142) in the DRY motif of the _1b-AR (Scheer et al., 1997) and is postulated to be involved in a proton shuttle during receptor activation. Because the net effect is an uptake of a proton from the solvent, this activating residue is favored under low cytosolic pH and the ultimate proton transfer by neutrality of D142. Because we find that the K331 _1b-AR mutations lead to a decreased acid strength, it is unlikely that D142 would account for the acidic pK_a change. In rhodopsin, mutation of E134 results in pK_a changes of the basic pH profile not the acidic pK_a (Cohen et al., 1993). Thus, changes in the acidic pK_a of the _1b-K331-AR mutations are likely attributable to D125.

However, as another test that changes in the pK_a of the K331 mutants is due to effects on D125, a salt-bridge switch mutant (D125K/K331D) was analyzed for its pH-dependent binding behavior. The K331D single mutant displays a higher pK_a value (Fig. 6, Table 1) as well as higher epinephrine binding affinity. The switch mutant, however, reversed the effects on both epinephrine affinity and pK_a to values seen for the _1b-D125A-AR single mutant (Fig. 6, Table 1). This suggests that D125 in the K331D receptor mutant is responsible for the increased pK_a. The lower binding affinity of epinephrine for the switch mutant is attributable to changes in the pharmacology of the ligand binding pocket. In unpublished studies, we have shown that the switch mutation displays 6-fold lower affinity for protonated amine agonists such as epinephrine, but a 6-fold higher affinity for acid derivatives of the catecholamine in which the positive charge of the catecholamine is replaced by a negative charge (Porter and Perez, 1999).

Various types of amino acid substitutions at K331 displayed similar phenotypes with higher epinephrine binding affinities and shifts in the pK_a for D125. Hydrophobicity, aromaticity, and electrostatic interactions might have expectantly given differential pK_a values. However, it does not seem that the type of substituted amino acid affected the ionization state of D125, suggesting that position 331 of the _1b-AR mutant is not influencing D125 after salt-bridge disruption. This result is consistent with the model of light-dependent proton pumping described for a related seven TMD receptor, bacteriorhodopsin. In this model, proton transfer includes the interactions of D85 and K216, which has the covalently bound all-trans-retinal forming the protonated Schiff base. These amino acid positions are conserved to D125 and K331 in the _1b-AR. Light-induced isomerization of the chromophore to the 13-cis configuration initiates proton transfer from the Schiff base to the D85 counterion (Otto et al., 1990). However, transient counterion interactions of the Schiff base with D212 after proton transfer to D85 have been postulated (Marti et al., 1991) suggesting that K216 of bacteriorhodopsin shifts away from D85 after proton transfer to stabilize with D212. In bacteriorhodopsin, D212 is located four amino acids (one helical turn) closer to the extracellular surface of the receptor than is K216 in TMD seven. It is noteworthy that in the _1b-AR there is analogous conservation of this residue, D327, which also is four amino acids from K331. Mutagenesis studies in bacteriorhodopsin speculate the importance of D212 in selective reisomerization of the chromophore and reprotonation of the Schiff base during the regeneration phase of bacteriorhodopsin's photocycle (Dunach et al., 1990; Rothschild et al., 1990). Therefore, analogous to bacteriorhodopsin, after disruption of the constraining _1b-AR salt-bridge, formation of transient ionic interactions for K331 with D327 on the _1b-AR may occur. This may account for the apparent insensitivity of D125 pK_a changes with the type of amino acid substitution at K331 of the _1b-AR.

Because acid dissociation constants are equilibrium constants, changes in the pK_a can be equated to free-energy differences between wt and K331 _1b-AR mutants. This value can be used to estimate the strength of the _1b-AR salt-bridge and the amount of free energy needed for disruption. The calculated average change in free energy of 0.99 ± 0.07 kcal/mol can be compared with other systems where ionic interactions have been shown to be important in contributing to protein structure (Table 1). For example, a salt-bridge between I16 and D194 in -chymotrypsin with an interaction energy of 2.6 kcal/mol has been shown to be important for maintaining the active site structure of this serine protease (Fersht, 1972). More recently, in the xylanase of Bacillus circulans a salt-bridge between D83, which is highly conserved among all Family G glycosidases, and R136 contributes approximately 2 kcal/mol to the structural stability of the folded protein (Joshi et al., 1997). Estimation of the salt-bridge energy has also been described for the rhodopsin receptor system. A free-energy value of 2.6 kcal/mol has been calculated for the opsin receptor salt-bridge (Cohen et al., 1993). However, when opsin, a rhodopsin receptor that lacks the covalently bound chromophore, does attach 11-cis-retinal via Schiff base linkage, there is a significant increase to 7 kcal/mol in the free-energy strength of the salt-bridge (Sakmar et al., 1989). It has been suggested that the lower stability of the opsin salt-bridge allows for the deprotonated K296 to attack the incoming 11-cis-retinaldehyde during the formation of the Schiff base. A stronger salt-bridge would make disruption and this subsequent covalent reaction less favorable. In a similar manner, we hypothesize that the weak _1b-AR salt-bridge is consistent with a mechanism whereby the free energy of epinephrine binding (7.5 kcal/mol based on a K_i of 3 µM) is favorable for disrupting the salt-bridge and initiating the process of receptor activation (i.e., too strong of a salt-bridge would not allow epinephrine docking).

Another contributing factor to the weaker _1b-AR salt-bridge strength compared with that of rhodopsin and other proteins is the difference in the dielectric constant of the binding pocket. In rhodopsin, 11-cis-retinal binds much deeper (22 Å) in the binding pocket (Stryer et al., 1982) when compared with the ₂-AR, which was found to bind ligands about 10 Å below the surface (Tota and Strader, 1990). This would suggest that ARs have a more hydrophilic binding pocket than rhodopsin, which would weaken a salt-bridge interaction through charge competition with water. In addition, the buried salt-bridges of chymotrypsin and xylanase would calculate to higher free energies when compared with the _1b-AR. However, a water-accessible salt-bridge interaction has been described for barnase (Horovitz et al., 1990). Stabilization energies of 1.25 and 0.98 kcal/mol for two arginine-aspartic acid salt-bridges in barnase are similar to our calculation for the _1b-AR.

Without structural information, we cannot discriminate in our experimental system that differences in the pK_a are directly attributable to salt-bridge breakage or merely caused by a receptor conformation induced by constitutive activation of the K331 receptor mutant, thereby changing the D125 microenvironment. In other unpublished studies (Porter and Perez, 1999), the switch mutation, _1b-D125K/K331D-AR reversed the constitutive activity of the single mutations also suggesting that these two residues are interacting in a salt-bridge. Because the switch mutant is not constitutively active and also displays a similar pK_a as the D125A mutant receptor, it seems that the K331 receptor mutant effects on D125 pK_a are attributable to disruption of the salt-bridge and not to the receptors constitutive phenotype. Therefore, the experimental data is consistent to a salt-bridge paradigm where breakage results in a weakened acid strength of D125. We are currently assessing other previously described constitutively active _1b-ARs to determine whether all such mutations cause similar changes in D125 pK_a or represent different activation paradigms. There is no evidence as of yet to suggest that salt-bridge disruption is a conserved feature among biogenic amine receptors. Mutation of a K305 in TMD seven of the ₂-AR does not lead to constitutive activation (JE Porter and DM Perez, unpublished data). However, this residue is located a helical turn closer to the extracellular surface of the ₂-AR than K331 in the _1b-AR. Recent work in the ₂-AR in which a histidine replaced D113 in TMD three and a cysteine replaced N312 in TMD seven, followed by chelation with zinc, did result in a constitutive phenotype (Elling et al., 1999). The altered distance between these two helices introduced by the zinc (mimicked by a broken salt-bridge) is thought responsible for the active state stabilization. This suggests that these two helices and/or residues are involved in the activation process. Given the closer phylogenetic relationship of ₁-ARs to rhodopsin, conservation of activation mechanisms may only occur in the ₁-AR. On the other hand, recent work in the -opioid receptor has generated constitutively active receptors with mutations at D128 in TMD three and Y308 in TMD seven, conserved analogously to the ₁-ARs (Befort et al., 1999). It has been proposed that these two residues interact in a hydrogen bonding network similar to our salt-bridge hypothesis. Therefore, it may be likely that analogous constraints between TMD three and seven exist in other GPCRs and may represent part of a universal activation mechanism.