Proc Natl Acad Sci U S A 108(46): 18684-18689

PMC: PMC3219117

PMID: 22031696

Activation mechanism of the <em>β</em><sub>2</sub>-adrenergic receptor

Results

Two crystal structures of agonist-bound β₂AR were available when we performed this study. In one, the receptor was stabilized in an active conformation by the binding of a G-protein-mimetic nanobody in addition to a potent agonist; this structure [the “active structure,” Protein Data Bank (PDB) entry 3P0G; ref. 2] closely resembles the crystal structure of meta II (active) rhodopsin (5). In the other agonist-bound β₂AR structure (PDB entry 3PDS; ref. 17), the receptor lacks an intracellular binding partner; this structure is almost identical to that of inactive, inverse-agonist-bound β₂AR (the “inactive structure,” PDB entry 2RH1; ref. 1).

These crystal structures suggest that, even though agonist binding likely increases the fraction of active receptors, most of the receptor population remains inactive in the absence of a G protein or G-protein-mimetic binding partner. In other words, the lifetime of the active state is shorter than that of the inactive state in the absence of the G protein. Accordingly, unbiased MD simulations starting from the active structure, with the cocrystallized agonist BI-167107 but without the nanobody, transitioned from the active to the inactive receptor conformation on timescales of several microseconds (17); the reverse transition has never been reported in unbiased simulations and has been shown experimentally to occur on substantially longer timescales (19).

In the present study, we performed 76 unbiased, all-atom MD simulations of β₂AR starting from the active structure, with the cocrystallized agonist but without the nanobody (Table S1); the receptor was embedded in a hydrated lipid bilayer, and the simulations ranged in duration from 2 to 50 μs. In 36 of these simulations, the receptor transitioned spontaneously from the active conformation to a conformation matching the inactive structure (see SI Text and Table S2). In each of these cases, the receptor, once settled into the inactive conformation, remained there for the remainder of the simulation. These simulations did not incorporate any prior knowledge of the crystallographically observed inactive conformation, so the fact that they reached it spontaneously provides an indication of their predictive power; nonetheless, the pathways traversed between these two states may be affected by inaccuracies in the underlying force field models. In nine additional simulations of the crystallized complex, including the nanobody, the receptor never adopted an inactive conformation (see SI Text and Table S1).

While this paper was under review, a crystal structure of active β₂AR in complex with a G protein was reported (3); the conformation of β₂AR in this structure is similar to that in the nanobody-bound active structure. Five additional simulations of β₂AR starting from its conformation in the β₂AR–G-protein complex structure produced results similar to those reported in the remainder of this paper (see SI Text and Tables S1 and S2).

By analyzing the transitions of the agonist-bound receptor from the active conformation to the inactive conformation—a process we call deactivation—we can infer properties of the mechanism of the reverse transition, receptor activation. Each deactivation pathway represents a possible activation pathway in reverse. More importantly, these simulations allow us to watch in action the machinery that links the conformation of the ligand-binding site to that of the G-protein-binding site.

Simulations Display an Intermediate G-Protein-Binding Site Conformation.

During the deactivation process, the receptor’s G-protein-binding site typically assumed three major, discrete conformations: active, inactive, and a previously unobserved intermediate (Fig. 1). For the initial part of each simulation, usually several hundred nanoseconds, the intracellular side of the receptor remained in a conformation matching that observed crystallographically in complex with the nanobody, even though the nanobody was absent from the simulation. This active conformation differs from that of the inactive receptor primarily through shifts and rotations of helices 5, 6, and 7; additionally, the intracellular end of helix 7 (including the conserved NPxxY motif, Asn322^–Tyr326^{) maintains a distorted helical configuration in the active conformation, with the side chain of Tyr326}^{separating helix 6 from helix 3. (Superscripts refer to Ballesteros–Weinstein residue numbering; ref.}20). Crystal structures of meta II (active) rhodopsin and of opsin (both with and without the G_α C-terminal helix bound) display a similar arrangement of helices 5, 6, and 7 (5, 21, 22), except that the intracellular end of helix 6 is shifted approximately 5 Å farther from the helical bundle in the active structure of β₂AR (and in this initial part of the β₂AR simulations).

Fig. 1.

The β₂AR G-protein-binding site adopts three major conformations: active, inactive, and a previously unobserved intermediate. (A) The distance between Arg131^{and Leu272}^C_α atoms (a measure of helix 6 displacement) and the rmsd of Asn322^–Cys327^{backbone atoms [a measure of helix 7 conformation; relative to the inactive structure (PDB entry 2RH1)] are plotted for a simulation in which an agonist-bound receptor, initially in the conformation of the active structure (PDB entry 3P0G) spontaneously transitioned to an inactive conformation. Circles represent simulation snapshots sampled every 6 ns. Diamonds correspond to the active (3P0G) and inactive (2RH1) crystal structures, and to the average value of these coordinates in a simulation of the inactive structure with the cocrystallized inverse agonist bound (2RH1-sim;}Table S1, condition R). (B) Representative structures of the three major conformational clusters, each superimposed on the inactive structure (light blue). Data are from simulation 11 (Table S2).

Next, the G-protein-binding site usually adopted an intermediate conformation substantially different from either crystallographic conformation. Upon transition to this intermediate, helix 7 assumes its inactive conformation, with its intracellular end adopting an undistorted α-helical conformation and moving away from helices 3 and 6, such that Tyr326^{loses contact with helix 6 (}Fig. 1). This helix 7 motion opens a space between the intracellular ends of helices 2, 3, 6, and 7, accommodating approximately five additional water molecules (Fig. S1). Helices 3, 5, and 6 maintain a more active-like conformation; in particular, helix 6 maintains its separation from helix 3 (Fig. 1). Helices 5 and 6 display substantial mobility within the intermediate, however, sometimes adopting a conformation where helix 6 moves approximately 7 Å in the direction of helix 5, which in turn moves approximately 5 Å toward helix 3 (Fig. S2).

After a typical residence time of several hundred nanoseconds to several microseconds in the intermediate conformation, the receptor shifted into the inactive conformation. The intracellular end of helix 6 moved inward, rotating such that the side chains of four residues (Met279^{, Leu275}^{, Leu272}^{, and Glu268}^{) previously separated from helix 3 by helix 5 now formed contacts with helix 3. The side chain of Tyr219}^{moved from between helices 3 and 6 to the lipid-exposed side of helix 6, adopting the dominant conformation seen in simulations of the inactive structure (}14) and allowing the Arg131^–Glu268^{ionic lock salt bridge to form (}Fig. S3). Despite the bound agonist, the conformations and dynamics of the G-protein-binding site following deactivation were nearly identical to those seen in inverse agonist-bound β₂AR simulations (14); the receptor interconverted between two distinct inactive conformations—a predominant one with the ionic lock formed and a minor one with it broken (Fig. S4).

The β₂AR contains several titratable residues, including Asp79^{, Asp130}^{, and His172}^{, for which protonation states are uncertain or may change during activation (}16, 24). We performed simulations with each of these residues, and combinations thereof, in different protonation states (Table S1). Although we found that simulations with Asp130^{charged typically reached the inactive state nearly twice as quickly as those with it neutral (}SI Text)—in accord with experimental observations indicating that Asp^{protonates upon activation of rhodopsin (}23) and likely of β₂AR as well (24)—the deactivation pathway was substantively affected only by the protonation state of Asp79^{. With Asp79}^{charged, deactivation consistently proceeded via the intermediate described above. With Asp79}^{neutral, the receptor followed this canonical pathway in three of four simulations that deactivated, but took an alternative pathway in the remaining one, suggesting that the two pathways may coexist in a certain pH range. Although broadly similar to the canonical pathway, the alternative pathway differs in that the initial motion of helix 7 is not to its inactive conformation; rather, it adopts an undistorted helical configuration in place, without moving away from helix 6 (}Fig. S5). As the deactivation process proceeds, helix 7 eventually relaxes to its inactive conformation, but only after helix 6 has transitioned to its own inactive conformation. The conformation of helix 7 in the intermediates on this pathway is similar to that observed in recent crystal structures of the agonist-bound A_2A adenosine receptor (25, 26). Several other simulations followed the initial steps of the alternative deactivation pathway but did not reach the inactive conformation (Fig. S5). Although we focus on simulations with Asp79^{charged in the following sections, the key results also hold for the alternative pathway.}

The Ligand-Binding and G-Protein-Binding Sites Appear to Be Loosely Coupled.

Although the differences in the ligand-binding site of the active and inactive structures are much more subtle than those in the G-protein-binding site, these small changes led Rasmussen et al. (2) to suggest an allosteric network connecting the two sites (Fig. 2A). The primary change in the ligand-binding site upon activation is a motion of helix 5 toward helix 6, with Ser207^{in particular moving approximately 1.7}Å to form a hydrogen bond with the agonist’s p-hydroxyl group. As a result, Ser207^{and Pro211}^{move into the space occupied by Ile121}^{in the inactive structure. Ile121}^{consequently shifts toward helix 6 into the space previously occupied by Phe282}^{, which in turn moves away from helix 3, tilting the intracellular end of helix 6 away from helix 3.}

Fig. 2.

The allosteric network underlying β₂AR activation involves three regions that each switch individually between discrete conformations. (A) Active structure of β₂AR, with three key regions highlighted. (B) Representative structures of the dominant conformations for each region in simulation. (C) Quantities indicative of each region’s conformation: the vectorial displacement of the Ser207^C_α atom away from helix 7 (see SI Text) and the distance between the Ser207^{side-chain oxygen and agonist}p-phenol oxygen atoms (ligand-binding site); the rmsd from the active and inactive structures of the nonsymmetric, non-hydrogen atoms in Ile121^{and Phe282}^{(connector); and the two quantities plotted in}Fig. 1A (G-protein-binding site). The time series are smoothed with a 9.9-ns running average. These quantities and related ones are used to classify the conformation of each region (see SI Text). The shaded bars below the plots indicate which conformation is adopted by each region as a function of time (active/inactive for the ligand-binding site and connector; active/intermediate/inactive for the G-protein-binding site), with color coding matching that of B. Data are from simulation 11 (Table S2).

Our simulations not only capture this allosteric network in action, but also indicate that it involves three distinct regions that can each switch individually between multiple conformational states. The residues in and just below the ligand-binding site transition from their active to their inactive conformation as the G-protein-binding site deactivates, but they do not all move in concert with one another or with the G-protein-binding site. Instead, they divide into two spatial clusters, with the residues in each tending to move together: one cluster consists of helix 5 residues in the ligand-binding site, near Ser207^{, whereas the other, which we term the connector region (or simply connector) given its location between the ligand-binding and G-protein-binding sites, comprises primarily Ile121}^{and Phe282}⁽Fig. 2B). Both regions adopt discrete conformational states which resemble either their active or inactive structures, although the connector acts more as a binary switch whereas the ligand-binding site displays more conformational diversity (Fig. 2B and C, and Fig. S6).

Although the ligand-binding site couples to the connector, and the connector to the G-protein-binding site, both couplings are loose (Fig. 3, Figs. S7 and S8, and Table S3). The ligand-binding site fluctuates between active and inactive conformations regardless of the conformations of the connector and the G-protein-binding site. The relative population of ligand-binding site conformations does depend on the connector conformation, however, with the ligand-binding site adopting an active conformation 78% and 11% of the time when the connector is in its active or inactive conformations, respectively, in our simulations (an important caveat is that these simulations do not represent equilibrium sampling). The connector can also switch between active and inactive conformations when the G-protein-binding site is in the active or intermediate conformations; in our simulations, the connector was in its active conformation 86%, 11%, and 0% of the time when the G-protein-binding site was active, intermediate, or inactive, respectively. In other words, the parts of the receptor appear to behave not like interlocked gears that move in synchrony, but rather like a set of loosely coupled switches, each of which influences the probability that its neighbor(s) will adopt one state or another.

Fig. 3.

The ligand- and G-protein-binding sites are loosely coupled, but an inactive G-protein-binding site restricts the connector to its inactive conformation. Each set of three horizontal bars indicates which conformation is adopted by the ligand-binding site, the connector, and the G-protein-binding site during that simulation. The upper inset shows which conformations are adopted in each major receptor state in our agonist-bound simulations; large text indicates a predominant conformation. Simulation numbers match those in Table S2. The final simulation was initiated from the inactive structure, with the inverse agonist carazolol bound (Table S1, condition R).

Activation May Begin near the G-Protein-Binding Site.

Despite the loose coupling between the connector and the G-protein-binding site, our simulations suggest that the inactive G-protein-binding site conformation precludes an active connector conformation. When the G-protein-binding site adopts active or intermediate conformations—both characterized by an outward shift of the intracellular end of helix 6—the connector can assume both active and inactive conformations, and often switches between them (Fig. 3 and Fig. S6). By contrast, the connector consistently adopts an inactive conformation before the G-protein-binding site does, and does not assume an active conformation once the G-protein-binding site has taken on the inactive conformation.

This behavior suggests that, during the reverse process of receptor activation, the G-protein-binding site likely leaves its inactive conformation before the connector assumes an active conformation. This order of events runs counter to a simple sequential model of activation in which agonist-induced structural changes in the ligand-binding site lead to changes in the connector, which only then lead to changes in the G-protein-binding site. Instead, our simulations suggest that the activation process typically begins at the G-protein-binding site, with the rest of the receptor only subsequently adopting an active conformation. Our simulations are not sufficiently long to capture a full transition from the inactive to the active receptor state, however, and activation pathways beginning at the ligand-binding site may also be possible.

If activation may begin on the intracellular side of the receptor, what role does agonist binding play in the activation process? Experimentally, agonists bind more tightly to the active state than the inactive state (27), and agonist binding thus shifts the ligand-binding site equilibrium between active and inactive conformations; free-energy-perturbation calculations confirm that this is the case in our simulations as well (see SI Text). Once the G-protein-binding site has adopted an intermediate conformation, the agonist thus stabilizes the active conformation of the connector, helping to maintain the G-protein-binding site in its intermediate or active conformations. Thus, the initial step of activation appears to depend on a preexisting conformational equilibrium in the G-protein-binding site, which the agonist helps bias toward active states, much along the lines proposed by Schwartz et al. (28).

Discussion

Our simulations of β₂AR transitioning from an active state to an inactive state allow us to build a picture of the machinery underlying the reverse process, receptor activation. The simulations show how agonist-induced changes in the conformation of the ligand-binding site can favor consolidation of activating conformational changes at the distant G-protein-binding site. Three functionally relevant regions of the receptor—the ligand-binding site, the G-protein-binding site, and a connector between them—appear to be only loosely coupled to one another. Each can take on multiple distinct conformations, and each can typically change conformation without a corresponding conformational change in the other two regions.

This rather soft coupling mechanism has several implications for signaling by drugs and endogenous ligands. It suggests that partial agonists need not induce distinct conformations of the ligand-binding site to achieve distinct levels of efficacy; instead, it suffices for them to differentially shift the equilibrium between active and inactive conformations of the ligand-binding site. Loose coupling also helps explain the difficulty of stabilizing β₂AR in a homogeneous active state with an agonist (29): Even an agonist that locks the ligand-binding site into an active conformation may not lock the entire receptor in its active state.

In our simulations, β₂AR also adopts a set of intermediate conformations on the pathway between the active and inactive states. Indeed, the ability of the ligand-binding site, the G-protein-binding site, and the connector to individually change conformation implies that many possible combinations—each of which represents a distinct receptor conformation—may be visited during activation or deactivation. The intermediate conformations of the G-protein-binding site (shown in Fig. 1 for the canonical pathway and in Fig. S3 for the alternative pathway) deserve special attention because of their potential implications for intracellular signaling. An extensive body of pharmacological evidence indicates that individual GPCRs are capable of adopting multiple conformations with distinct intracellular signaling profiles in response to binding of different ligands (6 –8). Fluorescence spectroscopy and ligand-binding affinity measurements have suggested that activation by full agonists involves at least one intermediate conformation (30, 31). The major intermediate conformations of the G-protein-binding site adopted in our simulations may well have a different signaling profile from the crystallographically observed active conformation and may be differentially stabilized by different ligands.

We speculate that, during the typical activation process, the G protein may first bind to the receptor when the receptor’s G-protein-binding site is in an intermediate conformation. The major intermediate of Fig. 1 has an even larger cleft between the intracellular ends of helices 2, 3, 5, 6, and 7 than the active conformation, and it could thus also accommodate the C-terminal α-helix of a G protein (Fig. S9). Binding of a G protein (or a G-protein mimetic) to this intermediate may promote its transition to the active conformation: In the presence of an appropriately positioned binding partner, the inward motion of the NPxxY region would allow burial of otherwise exposed hydrophobic surface area in helix 7. Indeed, spectroscopic studies of rhodopsin have suggested a two-step interaction mechanism in which GDP-bound transducin first forms a complex with an intermediate conformation of rhodopsin, and subsequently the two undergo a conformational change resulting in the fully active receptor conformation and GDP release from the G protein (32). An appropriate nanobody might stabilize the G-protein-binding site in the intermediate rather than active conformation, enabling crystallographic determination of this intermediate structure.

Our simulations suggest that receptor activation may occur as shown in Fig. 4. Once the G-protein-binding site has adopted an intermediate conformation—presumably due to fluctuation within a preexisting conformational equilibrium—the connector can adopt an active conformation. The agonist then serves to stabilize the ligand-binding site, and thus the connector, in an active conformation. Finally, perhaps as a result of G-protein binding, helix 7 changes configuration such that the G-protein-binding site adopts its active conformation.

Fig. 4.

A typical β₂AR activation pathway as inferred from our simulations. In A, an agonist has bound to an inactive receptor. (B) The activation process begins on the receptor’s intracellular side with an outward motion of helix 6, past Tyr219^{, bringing the receptor to an intermediate state. Here, the connector and the ligand-binding site are in equilibrium between inactive and active conformations; the bound agonist stabilizes the active conformations. (}C) A G protein may bind to the intermediate state, favoring the final step on the activation pathway: conformational change in the NPxxY motif (helix 7) that positions the side chain of Tyr326^{close to that of Tyr219}^{. Other activation pathways are also possible.}

Several caveats are worth noting. First, molecular dynamics simulations are subject to errors from several sources, so the actual activation mechanism might differ from that suggested by our simulations. Second, we did not simulate spontaneous dissociation of the nanobody, an event that should take place only on far longer timescales. The receptor might follow a different pathway between inactive and active states in the presence of a nanobody or G protein, particularly if such a binding partner is necessary to induce early conformational changes on the activation pathway (although fluorescence data suggests that activation-like conformational fluctuations in helix 6 of the agonist-bound receptor take place even in the absence of an intracellular binding partner; ref. 2). Finally, intracellular loop 3 (ICL3), which has been either absent from or unresolved in each of the available β₂AR structures, was also absent from our simulations. Removal of the bulk of ICL3 by partial tryptic digest does not appreciably alter receptor function (33), but ICL3 likely does interact with the G protein and might affect the activation pathway.

Rhodopsin activation has proved more amenable to experimental characterization than activation of other GPCRs. Although these experiments have not provided the same level of spatial and temporal detail as our simulations, rhodopsin activation involves similar conformational changes (5, 34) and also takes place via a set of intermediate states (35). In agreement with our results on β₂AR, FTIR spectroscopy on azido-labeled rhodopsin suggests that the first intracellular conformational changes involve helices 5 and 6 (36), and recent NMR data indicate that Tyr^{and particularly Tyr}^{play a critical role in rhodopsin activation (}37). On the other hand, rhodopsin has several early activation intermediates that appear to involve only small conformational changes near the binding pocket (35). That these intermediates are populated in an obligate, sequential order is suggestive of an “induced fit” activation process in which retinal isomerization causes conformational changes in the ligand-binding site that then propagate toward the intracellular side. It also suggests tighter coupling between receptor regions, consistent with the extremely low basal activity and extremely high photon detection efficiency of rhodopsin (38). These differences suggest some variability in signal transduction mechanisms among GPCRs, despite the involvement of highly conserved residues (39).

A promising frontier in drug development is the design of ligands that differentially modulate signaling pathways governed by a single GPCR, stimulating beneficial pathways while blocking deleterious ones (6 –8). Rational design of such “biased” or “functionally selective” ligands demands a structural understanding of the different conformations a GPCR can assume and of how each conformation influences various downstream G-protein and arrestin signaling pathways. Identifying the complete set of functionally relevant GPCR conformations and their implications for downstream signaling will require substantial additional work: These conformations may differ from one receptor to another, and some may lie off the activation pathway. By capturing in simulation an activation mechanism of an archetypal GPCR, however—both the major conformational intermediates and the machinery responsible for coupling the ligand- and G-protein-binding sites—our results may serve as a step toward the goal of creating drugs that provide finer control of GPCR signaling.

Methods

We performed all-atom molecular dynamics simulations of β₂AR, with lipids and water represented explicitly, using the CHARMM (Chemistry at Harvard Molecular Mechanics) force field (40) on Anton (18), a special-purpose machine that accelerates such simulations by orders of magnitude. No artificial (biasing) forces were applied. Further details are provided in SI Text.

^{D. E. Shaw Research, New York, NY 10036; and}

^{Center for Computational Biology and Bioinformatics, Columbia University, New York, NY 10032}

^{To whom correspondence may be addressed. E-mail:}moc.hcraeseRwahSED@rorD.noR or moc.hcraeseRwahSED@wahS.divaD.

Edited by Michael Levitt, Stanford University School of Medicine, Stanford, CA, and approved September 6, 2011 (received for review June 29, 2011)

Author contributions: R.O.D., D.H.A., P.M., A.C.P., D.W.B., and D.E.S. designed research; R.O.D., D.H.A., P.M., T.J.M., and H.X. performed research; R.O.D., D.H.A., P.M., T.J.M., and A.C.P. analyzed data; and R.O.D., D.H.A., P.M., T.J.M., A.C.P., D.W.B., and D.E.S. wrote the paper.

^{R.O.D., D.H.A., and P.M. contributed equally to this work.}

Edited by Michael Levitt, Stanford University School of Medicine, Stanford, CA, and approved September 6, 2011 (received for review June 29, 2011)

Freely available online through the PNAS open access option.

Abstract

A third of marketed drugs act by binding to a G-protein-coupled receptor (GPCR) and either triggering or preventing receptor activation. Although recent crystal structures have provided snapshots of both active and inactive functional states of GPCRs, these structures do not reveal the mechanism by which GPCRs transition between these states. Here we propose an activation mechanism for the β₂-adrenergic receptor, a prototypical GPCR, based on atomic-level simulations in which an agonist-bound receptor transitions spontaneously from the active to the inactive crystallographically observed conformation. A loosely coupled allosteric network, comprising three regions that can each switch individually between multiple distinct conformations, links small perturbations at the extracellular drug-binding site to large conformational changes at the intracellular G-protein-binding site. Our simulations also exhibit an intermediate that may represent a receptor conformation to which a G protein binds during activation, and suggest that the first structural changes during receptor activation often take place on the intracellular side of the receptor, far from the drug-binding site. By capturing this fundamental signaling process in atomic detail, our results may provide a foundation for the design of drugs that control receptor signaling more precisely by stabilizing specific receptor conformations.

Keywords: molecular dynamics, ligand, biased signaling, functional selectivity, beta-adrenergic

Abstract

G-protein-coupled receptors (GPCRs) represent the largest class of drug targets, and most drugs that act on GPCRs do so by either triggering or preventing receptor “activation,” which involves a transition from an inactive, nonsignaling receptor conformation to an active conformation that induces G-protein-mediated signaling. This conformational change allows the GPCR to transmit signals into the cell in response to the binding of drugs or endogenous ligands to the cell’s exterior. Crystallography has now revealed structures of two GPCRs, the β₂-adrenergic receptor (β₂AR) and rhodopsin, in both an inactive state (bound to an inverse agonist) and an active state (bound to both an agonist and a G protein or G-protein-mimetic intracellular binding partner) (1 –5).

Despite these fundamental advances, the mechanism by which GPCRs transition between inactive and active states remains unclear. What intermediate conformations does a receptor adopt en route to its active state? Do different regions of the receptor move in a concerted or an independent manner during activation? How do subtle agonist-induced conformational changes in the ligand-binding site lead to much larger conformational changes at the distant G-protein-binding site?

These questions are particularly important in light of pharmacological data suggesting that GPCRs can adopt multiple, distinct signaling states in response to the binding of different ligands (6 –8). Although the number of functionally distinct states and their conformations remain unknown, some are thought to be intermediates along the activation pathway (7, 9). The rational design of drugs that would bring about a particular, desired mode of GPCR signaling requires a better understanding of the different states accessible to each GPCR and of the mechanisms by which bound ligands can differentially induce or stabilize those states.

Here, we use molecular dynamics (MD) simulations to propose an atomically detailed activation mechanism for β₂AR, an archetypal GPCR. Although MD and related techniques have been utilized extensively to study conformational dynamics of GPCRs (e.g., refs. 10 –16), unbiased simulation of the transition from one crystallographically validated GPCR functional state to another has historically been infeasible, because the timescales of such transitions exceed those that have been accessible to MD. We recently demonstrated the feasibility of such simulations (17), using specialized computer hardware that dramatically accelerates MD computations (18); that study included two simulations in which agonist-bound β₂AR transitioned from an active to an inactive crystallographically observed conformation. In that paper, we used those simulations only to confirm that the energetically favored conformation of agonist-bound β₂AR in the absence of an intracellular binding partner matched the (inactive) conformation determined crystallographically (17). In the current paper, we use many such simulations, performed under several different conditions, to build a detailed picture of the molecular machinery, intermediate states, and conformational dynamics underlying β₂AR activation.

Click here to view.

Acknowledgments.

We thank Brian Kobilka, Morten Jensen, and Willy Wriggers for helpful discussions, Kim Palmo for assistance in evaluating force fields, and Rebecca Kastleman and Mollie Kirk for editorial assistance.

Acknowledgments.

Footnotes

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1110499108/-/DCSupplemental.

Footnotes

References

1. Cherezov V, et al High-resolution crystal structure of an engineered human β2-adrenergic G protein-coupled receptor. Science. 2007;318:1258–1265.[Google Scholar]
2. Rasmussen SGF, et al Structure of a nanobody-stabilized active state of the β₂ adrenoceptor. Nature. 2011;469:175–180.[Google Scholar]
3. Rasmussen SGF, et al Crystal structure of the β₂ adrenergic receptor-Gs protein complex. Nature. 201110.1038/nature10361. [Google Scholar]
4. Palczewski K, et al Crystal structure of rhodopsin: A G protein-coupled receptor. Science. 2000;289:739–745.[PubMed][Google Scholar]
5. Choe HW, et al Crystal structure of metarhodopsin II. Nature. 2011;471:651–655.[PubMed][Google Scholar]
6. Galandrin S, Oligny-Longpre G, Bouvier MThe evasive nature of drug efficacy: Implications for drug discovery. Trends Pharmacol Sci. 2007;28:423–430.[PubMed][Google Scholar]
7. Urban JD, et al Functional selectivity and classical concepts of quantitative pharmacology. J Pharmacol Exp Ther. 2007;320:1–13.[PubMed][Google Scholar]
8. Rajagopal S, Rajagopal K, Lefkowitz RJTeaching old receptors new tricks: Biasing seven-transmembrane receptors. Nat Rev Drug Discov. 2010;9:373–386.[Google Scholar]
9. Kobilka BK, Deupi XConformational complexity of G-protein-coupled receptors. Trends Pharmacol Sci. 2007;28:397–406.[PubMed][Google Scholar]
10. Saam J, Tajkhorshid E, Hayashi S, Schulten KMolecular dynamics investigation of primary photoinduced events in the activation of rhodopsin. Biophys J. 2002;83:3097–3112.[Google Scholar]
11. Shi L, et al β₂ Adrenergic receptor activation: Modulation of the proline kink in transmembrane 6 by a rotamer toggle switch. J Biol Chem. 2002;277:40989–40996.[PubMed][Google Scholar]
12. Grossfield A, Pitman MC, Feller SE, Soubias O, Gawrisch KInternal hydration increases during activation of the G-protein-coupled receptor rhodopsin. J Mol Biol. 2008;381:478–486.[Google Scholar]
13. Tikhonova IG, et al Atomistic insights into rhodopsin activation from a dynamic model. J Am Chem Soc. 2008;130:10141–10149.[Google Scholar]
14. Dror RO, et al Identification of two distinct inactive conformations of the β₂-adrenergic receptor reconciles structural and biochemical observations. Proc Natl Acad Sci USA. 2009;106:4689–4694.[Google Scholar]
15. Provasi D, Filizola MPutative active states of a prototypic G-protein coupled receptor from biased molecular dynamics. Biophys J. 2010;98:2347–2355.[Google Scholar]
16. Vanni S, Neri M, Tavernelli I, Rothlisberger UA conserved protonation-induced switch can trigger “ionic-lock” formation in adrenergic receptors. J Mol Biol. 2010;397:1339–1349.[PubMed][Google Scholar]
17. Rosenbaum DM, et al Structure and function of an irreversible agonist-β₂ adrenoceptor complex. Nature. 2011;469:236–240.[Google Scholar]
18. Shaw DE, et al. Millisecond-scale molecular dynamics simulations on Anton; Proceedings of the Conference on High Performance Computing, Networking, Storage and Analysis (SC09); New York: Assoc Computing Machinery; 2009. [PubMed]
19. Vilardaga JP, Bünemann M, Krasel C, Castro M, Lohse MJMeasurement of the millisecond activation switch of G protein-coupled receptors in living cells. Nat Biotechnol. 2003;21:807–812.[PubMed][Google Scholar]
20. Ballesteros JA, Weinstein HIntegrated methods for the construction of three dimensional models and computational probing of structure function relations in G protein-coupled receptors. Methods Neurosci. 1995;25:366–428.[PubMed][Google Scholar]
21. Scheerer P, et al Crystal structure of opsin in its G-protein-interacting conformation. Nature. 2008;455:497–502.[PubMed][Google Scholar]
22. Park JH, Scheerer P, Hofmann KP, Choe HW, Ernst OPCrystal structure of the ligand-free G-protein-coupled receptor opsin. Nature. 2008;454:183–187.[PubMed][Google Scholar]
23. Vogel R, et al Functional role of the “ionic lock”—an interhelical hydrogen-bond network in family A heptahelical receptors. J Mol Biol. 2008;380:648–655.[PubMed][Google Scholar]
24. Ghanouni P, et al The effect of pH on β₂ adrenoceptor function: Evidence for protonation-dependent activation. J Biol Chem. 2000;275:3121–3127.[PubMed][Google Scholar]
25. Xu F, et al Structure of an agonist-bound human A_2A adenosine receptor. Science. 2011;332:322–327.[Google Scholar]
26. Lebon G, et al Agonist-bound adenosine A_2A receptor structures reveal common features of GPCR activation. Nature. 2011;474:521–525.[Google Scholar]
27. Kent RS, De Lean A, Lefkowitz RJA quantitative analysis of beta-adrenergic receptor interactions: Resolution of high and low affinity states of the receptor by computer modeling of ligand binding data. Mol Pharmacol. 1980;17:14–23.[PubMed][Google Scholar]
28. Schwartz TW, Frimurer TM, Holst B, Rosenkilde MM, Elling CEMolecular mechanism of 7TM receptor activation—a global toggle switch model. Annu Rev Pharmacol Toxicol. 2006;46:481–519.[PubMed][Google Scholar]
29. Ghanouni P, et al Functionally different agonists induce distinct conformations in the G protein coupling domain of the β₂ adrenergic receptor. J Biol Chem. 2001;276:24433–24436.[PubMed][Google Scholar]
30. Swaminath G, et al Sequential binding of agonists to the β₂ adrenoceptor: Kinetic evidence for intermediate conformational states. J Biol Chem. 2004;279:686–691.[PubMed][Google Scholar]
31. Liapakis G, Chan WC, Papadokostaki M, Javitch JASynergistic contributions of the functional groups of epinephrine to its affinity and efficacy at the β₂ adrenergic receptor. Mol Pharmacol. 2004;65:1181–1190.[PubMed][Google Scholar]
32. Shichida Y, Morizumi TMechanism of G-protein activation by rhodopsin. Photochem Photobiol. 2007;83:70–77.[PubMed][Google Scholar]
33. Rubenstein RC, Wong SK, Ross EMThe hydrophobic tryptic core of the β-adrenergic receptor retains Gs regulatory activity in response to agonists and thiols. J Biol Chem. 1987;262:16655–16662.[PubMed][Google Scholar]
34. Altenbach C, Kusnetzow AK, Ernst OP, Hofmann KP, Hubbell WLHigh-resolution distance mapping in rhodopsin reveals the pattern of helix movement due to activation. Proc Natl Acad Sci USA. 2008;105:7439–7444.[Google Scholar]
35. Park PSH, Lodowski DT, Palczewski KActivation of G protein-coupled receptors: Beyond two-state models and tertiary conformational changes. Annu Rev Pharmacol Toxicol. 2008;48:107–141.[Google Scholar]
36. Ye S, et al Tracking G-protein-coupled receptor activation using genetically encoded infrared probes. Nature. 2010;464:1386–1389.[PubMed][Google Scholar]
37. Goncalves JA, et al Highly conserved tyrosine stabilizes the active state of rhodopsin. Proc Natl Acad Sci USA. 2010;107:19861–19866.[Google Scholar]
38. Smith SOStructure and activation of the visual pigment rhodopsin. Annu Rev Biophys. 2010;39:309–328.[PubMed][Google Scholar]
39. Nygaard R, Frimurer TM, Holst B, Rosenkilde MM, Schwartz TWLigand binding and micro-switches in 7TM receptor structures. Trends Pharmacol Sci. 2009;30:249–259.[PubMed][Google Scholar]
40. MacKerell AD, Jr, et al All-atom empirical potential for molecular modeling and dynamics studies of proteins. J Phys Chem B. 1998;102:3586–3616.[PubMed][Google Scholar]