We have developed and tested a rapid protocol for identifying amino acid residues that, when altered, affect the chemical environment of a ligand bound to a receptor. The method uses fluorescence activated cell sorting to screen libraries of randomly mutated receptors for mutations that alter the emission wavelengths of bound fluorescent ligand analogs. It is selective for mutations that preserve high affinity ligand binding at the cell surface, thereby minimizing or eliminating confounding classes of mutations such as those that alter yields of receptor synthesis, folding, or subcellular targeting. The usefulness of the approach was evaluated in two separate screens focusing on the different interactions of the yeast α-pheromone receptor with the bound agonist and antagonist.
Mutations affecting the fluorescence emission of receptor-bound agonist all led to red-shifting of the spectrum, indicative of increased polarity of the fluorophore environment. They were all located at sites predicted to reside at the interface between the predicted transmembrane region of Ste2p and its extracellular surface. Several of the substitutions were at amino acid residues that had previously been identified as likely sites of interaction with α-factor, including: 1) S47R and F55S at the extracellular end of TM1. A previous genetic analysis implicated residues S47 and T48 of Ste2p in direct interactions with the Gln10
of α-factor 14
. Furthermore, crosslinking has been observed between α-factor derivatized at Tyr13
and regions of Ste2p encompassing residues F55-R58 13
and between the unnatural amino acid p-benzoyl-L-phenylalanine incorporated at position 55 in Ste2p and α-factor 31
. 2) F204I, F204L, N205H, and N205Y at the extracellular ends of TM5 and TM6. Mutational approaches and crosslinking have previously implicated residues F204, N205, and Y266 in interactions with the N-terminal region of the ligand 15; 16; 25; 32
. Residue Y266 has been implicated in interactions with ligand 15; 16
. Although mutations at Y266 were not recovered from the screen, presumably because of low levels of cell surface expression, the site-directed substitution at Y266C resulted in a significant emission shift of bound [Lys7
]α-factor. 3) D275E and D275G in the short extracellular loop. Mutations at D275 have previously been reported to affect ligand binding and receptor activation 22
The screen also led to the recovery of mutations at particular residues that have not previously been implicated directly as sites of ligand interaction, even though they reside in known interacting regions. These sites include amino acids L44, N46, Q51, and I53, located in a region of TM1 in which other ligand-interacting residues have been identified. Similarly, mutations that we recovered at residues S207, Q272, G273, and T274 are located in the third extracellular loop, which mutagenic and crosslinking studies have previously implicated in interactions with ligand 15; 16; 17; 25; 32
Additional mutations affecting the fluorescence of bound agonist were recovered at the extracellular ends of TM2 (at F99, Y101, and L102) and TM3 (at Y128) in the predicted first extracellular loop of Ste2p, which has not previously been identified as a site of contact with ligand. A previous cysteine-scanning study of residues 100–135 in the first extracellular loop of Ste2p uncovered no substitutions that affected ligand binding affinity 33
, consistent with the binding data presented in 33; 34
. Evidence the recovered mutations in the first extracellular loop alter the emission of bound [Lys7
]α-factor through direct interactions indicative proximity of these residues to ligand, is provided by the following observations: 1) In accordance with the screening paradigm, these substitutions do not significantly alter affinities for α-factor. In addition, most of them also do not alter receptor signaling function, making it unlikely that they alter receptor conformation. 2) We identified additional site-directed substitutions at positions F99, Y101, and Y128 that do, in fact, alter ligand binding (see ). 3) The mutations in the first extracellular loop are the only substitutions recovered in the screen that are not located at, or near, previously-known sites of receptor-ligand interactions. 4) None of the sites of mutations recovered in the agonist screen reside at positions that are deep in the predicted transmembrane regions or at the intracellular surfaces of the receptor. If the mutations could induce changes in emission of NBD-labeled agonist via indirect effects (as was observed for the antagonist screen, see below) the recovered set of alleles would have been expected to include mutations at distant sites.
Three possible mechanisms for direct effects of mutations on the fluorescence of bound ligand all involve short distances between the altered residue and the ligand binding site: 1) Direct physical contact between the site of the mutation and the NBD-group of the bound ligand or its immediate site of attachment to ligand. Such contact could only occur if the mutated residue is within a few Å of the fluorophore or the linker. 2) Alteration of the electrostatic or dielectric environment of the bound fluorophores by substitutions at amino acid residues that are not in direct contact with fluorophore. The range of interactions capable of altering fluorescence emission of the bound ligand should be comparable to the range over which electrostatic effects are found to affect pKa
s of ionizable groups in proteins, generally less than about 10 Å 35; 36
. 3) Alteration of the accessibility of the fluorophores to solvent. In structures of GPCRs in complex with small molecule ligands 2; 6; 7; 37; 38
, the ligands are bound in cavities facing the extracellular surfaces of receptors. If this is also the case for peptide ligands, mutations affecting accessibility to solvent of groups on the ligand would be most likely to occur on surfaces lining these cavities.
The fluorescence changes observed on binding of [Lys7
]α-factor to receptors 19; 20
, together with the fact that the screen was only able to recover mutations resulting in red-shifting of the NBD emission (indicative of increasing polarity) suggests that the side chain of Lys7
resides near a non-polar region when bound to wild-type Ste2p. Further indications of the polarity of the environments of different regions of bound ligand can be derived from comparisons of the behaviors of the two differentially labeled agonists. Mutations N205H, Q272P, G273S, and D275E in the predicted second and third extracellular loops of the receptor result in increased [Dap3
]α-factor emission intensity compared with normal receptors (indicative of decreased polarity) whereas the same mutations cause decreased emission and red-shifting of bound [Lys7
]α-factor (indicative of increased polarity, perhaps accompanied by a decrease in the number of surface-exposed receptors). Together with additional differential interactions between particular mutated residues and the Dap3
positions on the ligand listed in , these observations support the existence of specific interactions between the substituted residues and the affected fluorophores and argue that the changes in emission are not due to mutation-induced global conformational changes in the receptor.
The environment of NBD in bound antagonists appears to differ significantly from that of NBD in bound agonist. Three different antagonists, all labeled with NBD at the equivalent of position Lys7 of normal α-factor, exhibit emission spectra upon binding to receptor that are considerably red-shifted (indicative of a shift to a polar environment) compared to the spectrum of similarly-labeled bound agonist (). Furthermore, in sharp contrast to the mutations affecting agonist fluorescence, the amino acid substitutions identified as causing changes in the fluorescence emission of receptor-bound antagonist all caused blue-shifting of the spectrum of NBD-labeled ligand, indicative of decreased polarity of the fluorophore environment. These changes in fluorescence appear to result from overall changes in the receptor conformation based on the following:
- Most of the recovered mutations affecting [dTyr3,Lys7(NBD),Nle12]α-factor fluorescence were located at positions in the protein that are unlikely to interact directly with ligand, based on the predicted seven-transmembrane segment topology. This suggests that they are affecting fluorescence via a mechanism that does not involve direct contact with ligand.
- None of the tested mutations that affect the emission of bound agonist had any significant effect on the emission spectrum of bound antagonist (Supplementary Table 3). Thus, either the NBD group of receptor-bound antagonist [dTyr3,Lys7(NBD),Nle12]α-factor does not interact with the regions of the receptor that appear to be in close proximity with agonist, or the NBD group of the antagonist resides in a highly solvent-exposed environment that can not be directly altered by mutations in the receptor.
- None of the mutations that alter the emission spectrum of bound Lys7-labeled antagonist had any effect on the fluorophore of receptor-bound Lys7-labeled agonist (Supplementary Table 4). This implies either that these mutation-induced conformational changes do not alter the environment of bound agonist, or that these mutations do not alter the conformation of agonist-bound states.
- All of the recovered mutations that alter the emission spectrum of receptor-bound antagonist [dTyr3,Lys7(NBD),Nle12]α-factor also confer on the mutant receptors the ability to signal in response to binding of ligands that act as antagonists of normal receptors (). In addition, most of the mutations affecting the [dTyr3,Lys7(NBD),Nle12]α-factor fluorescence also conferred constitutive signaling activity on the receptor. In contrast, none of the mutations affecting the emission of bound agonist resulted in significant (greater than 2-fold) constitutive activation of signaling with the exception of the I53F substitution, which exhibited an activation of 3.1 (± 0.8)-fold over normal receptors.
- All but one of the mutations that alter the fluorescence emission of bound antagonist α-factor also resulted in enhanced binding affinity for antagonist (). In addition, some of these mutations also increased affinity for agonist (Supplementary Table 4). Thus, mutations of the receptor that directly alter the environment of the Lys7 position of antagonists without inducing wider conformational rearrangements that affect ligand affinity must be rare or non-existent. In contrast, receptor mutations affecting the fluorescence of bound agonist generally did not affect ligand binding affinity ().
When compared to the native agonist, α-factor, all known antagonists toward the α-factor receptor contain alterations in the N-terminal region of the peptide, such as deletion of extreme N-terminal amino acids, while maintaining the same C-terminal region as α-factor. Thus, the C-terminal region appears to mediate initial binding to receptor while the N-terminal region provides interactions necessary for receptor activation 26; 28
Viewed in this context, mutations that blue-shift the emission of bound antagonists may be seen as restoring agonist-type interactions with the N-terminal of the antagonist, driving the receptor into the conformation that it would normally have when bound to agonist and causing the antagonist to function as a weak agonist. Consistent with this, most such mutations enhance the receptor’s overall binding affinity for both antagonist and some increase the affinity for agonist (; Supplementary Table 4
As a test of the idea that mutation-induced conformational shifts of the receptor moves the NBD fluorophore of bound antagonist into an environment similar to that of the fluorphore of bound agonist, we combined the mutations Q51H and D275E (which shift the fluorescence of bound agonist but do not, by themselves, affect the emission of bound antagonist) into alleles that also contain the constitutively activating mutation P258S (which does not affect the emission of bound agonist but shifts the emission of bound antagonist). The presence of the Q51H or D275E mutations in the same allele as the P258S substitution red-shifts the emission of the bound antagonist [dTyr3,Lys7(NBD),Nle12]α-factor compared to alleles containing the constitutive P258S mutation alone (). Thus, once the labeled region of the antagonist is induced (by the constitutively activating mutations) to interacting with receptor, its emission can be shifted by the same mutations that shift the emission of bound agonist.
The mutations recovered at positions M54 and F55 in the first transmembrane segment of Ste2p exhibit several properties indicative of direct interactions of these amino acids with N-terminal regions of the ligand that determine its efficacy: 1) The effects of mutations at these positions are ligand-specific: The F55S substitution reduces binding of both the antagonist [d
]α-factor and the agonist [Dap3
] α-factor to undetectable levels without significantly diminishing binding or signaling responses of the normal agonist [Lys7
] α-factor. 2) The F55L substitution is the only identified mutation that provides a spectral shift in the emission of bound antagonist without enhancing overall binding affinity or the EC50
of the antagonist. 3) The M54I and F55L substitutions shift the fluorescence emission of antagonist and enhance the maximal signaling responses to antagonists without causing any significant enhancement of constitutive signaling. These findings indicative of specific interactions between the first transmembrane segment of the receptor and the different N-terminal regions of the tested series of ligands observations are difficult to reconcile with previous evidence for direct contact between the extracellular regions of transmembrane segments 5–7 of the receptor and the N-terminal region of the ligand 26; 28
. Taken together, the current and previous results suggest either that the N-terminal of the ligand interacts simultaneously with both the first transmembrane segment and segments 5–7 of the receptor, or that the mutations at M54 and F55 exert their effects via indirect effects on the conformation of nearby helices in Ste2p.
Overall analysis of the binding and signaling responses of normal and mutant receptors to agonists and antagonists is consistent with the mechanistic model of receptor action shown in . In the figure, the conformational change associated with activation is schematically represented as a close approach of two regions of the receptor that allows both the N- and C-termini of the agonist to make separate interactions with different portions of the ligand binding site (). Under all conditions, the receptor undergoes transitions between the activated and un-activated states, however the relative populations of receptors in the different states is altered by ligands and mutations. The interaction between the C-terminal of the ligand and the receptor is the major contributor to initial binding. Weaker, but highly specific, binding between the activated-receptor conformation and the N-termini of agonists is required to maintain the receptor in the activated state (). (If the ligand is held in position by binding to the C-terminal, avidity effects could allow the interaction with the N-terminal to be quite weak.) The interaction with the N-terminal region of the ligand stabilizes the active state of the receptor (close approach of two regions of the receptor in the model). When interactions with both the N- and C-termini of ligands occur, a fluorophore attached to Lys7 of α-factor is brought into a non-polar environment at the receptor-ligand interface, causing a blue-shift in emission. Mutations at the boundary between the predicted aqueous and transmembrane regions of each extracellular loop are capable of red-shifting the emission of bound ligand by increasing the polar nature of the environment of bound agonist ().
Figure 4 Model for the binding and activation of normal and mutant α-factor receptors by different ligands. Ligand-interacting surfaces on the receptor are indicated by cross-hatching. Site I interacts with the C-termini of ligands and is likely to consist (more ...)
Viewed in the framework of the model, binding of true neutral antagonists to normal receptors would involve only the component of the receptor-agonist interactions involving the C-terminal of the ligand, leaving the fluorophore attached to Lys7 exposed to solvent and insensitive to effects of mutations that alter the environment of bound agonist (). The loss of activating interactions of the N-terminal regions of antagonists with receptors may be overcome by mutations such as M54I and F55L in Ste2p that could directly enhance interactions between the N-terminal portions of agonists and the activated state of the receptor (). Such mutations can alter the relative affinity of the ligand for the activated vs. un-activated states, converting a neutral antagonist to an agonist. Such a conversion could occur without any enhancement of the overall binding affinity for the ligand, as is observed for the F55L substitution.
Most of the mutations recovered from the antagonist screen provide simultaneous shifting of the emission spectrum, constitutive activation of receptors, enhanced overall binding affinity for ligands, and ability of receptors to be activated by ligands that serve as antagonists toward normal receptors ( and Supplementary Table 4
). Based on the specific locations of these mutations, and their degree of dispersion throughout the sequence and predicted topology of the receptor, it is unlikely that they all could exert their affects by directly altering receptor-ligand interactions. Instead, it seems more likely that they act indirectly by altering the equilibrium of the receptor between activated and un-activated states as shown in . Such an alteration would not be expected to alter the signaling responses to a true neutral antagonist, which, by definition, binds both states with equal affinity. Thus, the observed effects are most likely a result of the weak partial agonist activity of the antagonistic α-factor analogs, which has been reported previously 30
, and can be seen in the dose response curves presented in Supplementary Fig. 5b
and the notes to . Increased population of the activated state (which preferentially interacts with agonists), enhances the binding affinity of partial agonists for receptors, in addition to promoting constitutive signaling activity. Such behavior is consistent with an analytical expression describing the two-state model for receptor signaling (see 39
), which predicts that constitutive signaling and signaling responses to weak partial agonists will exhibit similar dependencies on the equilibrium constant relating the populations of activated and un-activated ligand-free receptors (see supplementary text
). However, it is surprising that substantial changes in the fluorescence emission of bound antagonist/partial agonist are observed for mutants that cause only small shifts in the population of activated receptors, as indicated by relatively weak constitutive activity. This suggests that, upon binding the weak partial antagonist, the mutant receptors may be efficiently converted to a conformation that is actually different from the normal antagonist-stabilized activated state. This new state could be intermediate between the normal activated and un-activated states, providing significant alteration of the environment of the bound ligand, but only inefficient activation of G protein.
The fluorescence-based screen described here provides a relatively rapid procedure for identifying receptor-ligand interactions that can complement existing biochemical approaches. The selectivity of the screen for mutant alleles that retain correct folding, efficient targeting to the cell surface, and high affinity ligand binding makes it more direct and specific than genetic approaches based on detection of loss of ligand binding or signaling. Since many mammalian GPCRs and other receptors can be expressed in functional form in yeast 10; 11
, and since fluorescent ligand analogs are available for a variety of receptors 40
, these procedures should be useful for characterizing ligand binding sites on a range of different receptors. Differential application of this approach to agonist and antagonist binding to the yeast α-factor receptor has provided a view of the distinct interactions of the two classes of ligands with receptors, leading to a mechanistic model for receptor responses to the different types of ligands.