Nano‐G_s Protein Peptidomimetics: Rational Design of Gα C‐Terminus‐Derived Peptides Mimicking Key Components of G_s‐β₂AR Interactions

ABSTRACT

G protein‐coupled receptors (GPCRs) are involved in most human physiological processes and one of the largest families of approved drug‐targeted proteins. Heterotrimeric Gαβɣ proteins bind to the intracellular cavity of the activated receptor mainly through the C‐terminal α5 helix of the Gα subunit (GαCT). Modulation of GPCR activity through intracellular GPCR binding sites is emerging. Here we develop highly active Gα_sCT‐derived peptidomimetics that stabilize the β₂ adrenergic receptor (β₂AR) in an active‐like conformation when the helical conformation of Gα_sCT is preserved by a covalent tether (“staple”). By rational design, integrating the information of two crystal structures showing different binding modes of Gα_sCT with β₂AR, an appropriate staple position was identified. The key interactions observed in the two distinct β₂AR structures (β₂AR‐G_s(empty)) and β₂AR‐T4L‐G_sCT‐CC), namely Y391 and E392, were substituted with non‐canonical amino acids to improve activity. Optimization resulted in the identification of a potent peptidomimetic capable of stabilizing an active‐like receptor conformation, whilst blocking receptor‐mediated cAMP formation. Molecular dynamics simulations indicated a peptidomimetic binding mode that may represent another intermediate state preceding that of β₂AR‐G_s(empty). We envision this approach to be useful for further structural and functional exploration of other GPCRs or as a tool in drug discovery.

G protein‐coupled receptors (GPCRs) are major therapeutic targets. Modulating GPCR activity through intracellular sites is evolving. A structure‐ and computation‐assisted approach generated small G protein‐derived peptidomimetics targeting the intracellular binding crevice of the β₂ adrenergic receptor mimicking features of the full G protein. Our approach is useful for structural and functional exploration of GPCRs and in drug discovery.

1. Introduction

G protein‐coupled receptors (GPCRs) constitute the largest family of integral membrane‐bound proteins in the human genome and are key proteins for cellular function. Upon activation by extracellular signal molecules such as ions, hormones, neurotransmitters, or other stimuli, GPCRs transduce these signals to elicit spatially and temporally finely tuned cellular responses through interaction with intracellular macromolecules such as canonical heterotrimeric guanosine triphosphate‐binding proteins (Gαβɣ proteins), kinases and arrestins [1, 2]. G proteins, assembled by each one of 18 Gα, 5 Gβ, and 12 Gɣ protein subunits, elicit a unique signaling response through different intracellular pathways mainly by the receptor's preference toward one or more of the Gα_s, Gα_i/o/t, Gα_q, and Gα₁₂ classes [3, 4]. GPCRs continue to be an important class of proteins for drug development [5], and the majority of GPCR modulators including those used in the clinic are molecules targeting extracellular orthosteric or allosteric binding sites [6].

Alternative approaches, receiving increasing attention, are to modulate specific GPCR downstream signaling by interaction of molecules with the receptor crevice that interacts with intracellular proteins. Besides pepducins, which are lipidated peptides mimicking the intracellular loops (ICL) of GPCR [7, 8, 9, 10, 11, 12], small molecules overlapping with the G protein binding site have been reported for the β₂ adrenergic receptor (β₂AR), and the chemokine receptors 2 and 9 [13, 14, 15, 16]. Proteinogenic peptides derived from the α5 C‐terminus of G proteins (GαCT) were initially reported by Hamm and coworkers as GPCR modulators binding to the intracellular receptor crevice to stabilize an active‐like receptor conformation and prevent G protein coupling [17, 18, 19, 20]. Similar peptide derivatives have proven useful for stabilizing, detecting and purifying receptors in their active conformations and as pharmacological tool compounds for studying signal transduction [18, 21, 22, 23, 24]. In addition, active conformation‐stabilizing peptidomimetics could provide new opportunities for drug discovery as modulators that preferentially recognize and bind (malfunctioning) active‐like state receptors [25]. We previously reported efforts to develop peptidomimetics (termed peptides from hereon) based on the CDR3 loop of nanobody 80 [26] (Nb80), a surrogate G_s protein mimetic that stabilizes the β₂AR in an active conformation [27], and by an (i, i+4)‐stapling approach of a peptide comprising the 15 C‐terminal amino acids of the Gα_s subunit (Gα_sCT₁₅) [28]. However, only weak effects of the peptides for inhibiting agonist‐induced cAMP signaling were observed and no signs of active‐like conformation‐state stabilization by the peptides were identified using these two approaches. More recently, Ballet and coworkers reported more efficacious (i, i+4)‐stapled peptides capable of stabilizing active states of the β₂AR and the dopamine‐1 receptor [23].

In this work, we developed a strategy to design peptides interfering with G_s‐β₂AR complex formation. We speculated it would be possible to further improve binding of the Gα_sCT₁₅ peptide to the β₂AR by introducing a longer staple spanning beyond the typically (i, i+4)‐stapled peptides. Here, we explore stapling over two helical turns using a (i, i+7)‐staple as a strategy to reduce the entropic cost during peptide‐receptor complex formation, using methodology previously developed by us [29]. Using CD and NMR/MD analysis we and others have reported that such (i, i+7)‐stapled peptides adopt an extended α‐helical conformation [29, 30]. Further, we employed computational approaches to guide peptide design and optimize receptor interactions by modifying the Gα_sCT₁₅ sequence with canonical and noncanonical amino acids that would strengthen proposed key interactions with the receptor. We identify potent peptides capable of stabilizing the β₂AR in active‐like conformation and blocking β₂AR‐mediated cAMP formation and determine the receptor binding mode by molecular dynamics simulations.

2. Results and Discussion

2.1. Basis for Design of a Gα_sCT₁₅‐Peptide as a Gα_s Protein Surrogate

For the rational design of a Gα_sCT₁₅ peptide as a Gα_s surrogate to the β₂AR, we reviewed the available high‐resolution structural information of G protein and receptor complexes with respect to the C‐terminal part of α5 (PDB ID: 3SN6). The first structure of a ternary receptor‐G protein complex obtained from a nucleotide free β₂AR‐G_s complex [31] showed major outward tilt of transmembrane domain 6 (TM6), which opens up a binding crevice to which the tip of the α5‐helix binds. In the β₂AR‐G_s(empty) complex, α5 forms the major interaction site of G_s with the receptor. A large hydrophobic patch is formed between Gα_sCT with TM5 and TM6 of the receptor. At the floor of the binding cavity, a cation‐π interaction is formed from Y391 to R131^3.50 from the preserved E(D)RY motif. This key interaction is also observed in many other receptor‐G_s(empty) complex structures and was thus considered as an important element of our design strategy [32] Interestingly, in a structure capturing the β₂AR in a complex with the carboxyl terminal 14 amino acids from Gα_s (β₂AR‐T4L‐G_sCT‐CC, PBD ID: 6E67) [33], Gα_sCT interacts differently with the receptor, but still adopts an α‐helical conformation. Gα_sCT is, however, rotated when compared to β₂AR‐G_s(empty) and instead of Y391 the neighboring E392 of the Gα_sCT interacts with R131^3.50 (Figure 1A,B). Notably, E392 is solvent exposed in the crystal structure of G_s(GDP) while Y391 is buried. The interaction observed in β₂AR‐T4L‐G_sCT‐CC was accordingly suggested to represent an initial binding mode between the β₂AR and GDP‐bound G_s protein (β₂AR‐G_s(GDP)), before the major structural rearrangements of α5 lead to solvent exposure of Y391 [33]. In agreement, MD simulations have shown that G_s(GDP) spontaneously binds to an extended interface of the activated β₂AR with an α‐helical conformation of α5 that reassembles the key interactions of the β₂AR‐T4L‐G_sCT‐CC structure and induces long‐range allosteric structural changes in the nucleotide binding pocket to initiate GDP release [34]. While the C‐terminal part of α5 is not resolved in any structure of inactive GDP bound G_i/t proteins, it adopts a helical conformation in the GDP bound inactive G_s protein [33]. This suggests that the G_s α5 C‐terminus can be recognized by and bind to the receptor in an α‐helical and not the unfolded conformation. In addition, secondary structure analysis using CD spectrometry shows that a Gα_sCT₁₅ peptide, as a separate entity, adopts a disordered conformation corresponding to that of a random coil [28].

FIGURE 1.

Design strategy. (A+B) Interactions of Gα_sCT as observed in β₂AR‐G_s(empty) (PDB ID: 3SN6) and β₂AR‐T4L‐G_sCT‐CC (PDB ID: 6E67) with receptor (green) and peptide (orange) shown as cartoons, highlighting key interactions that are targeted for optimization (shown as sticks). In β₂AR‐G_s(empty) Y391 forms a cation–π interaction with R131^3.50 (min. distance: 4.0 Å) while E392 is too distant to interact (min. distance: 7.2 Å). In β₂AR‐T4L‐G_sCT‐CC E392 forms a salt‐bridge with R131^3.50 (distance: 2.2 Å) while Y391 is too distant to interact (min. distance: 11.4 Å). (C) Root mean square fluctuation (RMSF) obtained from MD simulations of unbound and receptor‐bound peptides highlighting the intrinsic flexibility of each amino acid of the peptide (MD‐simulation data taken from [33]). Gray points are simulations of unbound peptide, violet points are simulations of β₂AR‐T4L‐G_sCT‐CC, and teal points are of Gα_sCT peptide bound to β₂AR from β₂AR‐G_s(empty) after deletion of the remainder of G_s. (D) Solvent‐exposed basic residues (shown as sticks and surface representation) at the intracellular receptor surface and (E) intracellular receptor surface colored by hydrophobicity [35] (blue and red referring to low and high hydropathy index, respectively) for indication of basic and hydrophobic patches to which the modified peptide is designed to bind with high affinity.

Based on this initial analysis, we hypothesized that constraining Gα_sCT in an α‐helical conformation could help to mimic the α5 of receptor‐bound full‐length G_s and at the same time reduce the entropic cost of binding‐induced folding of native Gα_sCT₁₅. Notably, however, time‐resolved structural mass spectrometry techniques revealed, that α5 retains some flexibility during the coupling process and only becomes fully stabilized after prolonged incubation with apyrase yielding the nucleotide free complex [31, 36]. Thus, a staple should accordingly be designed in a way that would enhance formation of an α‐helix, while keeping some flexibility at both termini to allow the peptide to conformationally adapt to the receptor during the binding process. Moreover, the staple should be introduced at a position that would allow interaction of Y391 or E392 with the intracellular binding pocket of the receptor, and at sites allowing energetically favorable interactions of the peptide with the hydrophobic patch at TM5/TM6. To further strengthen receptor‐peptide interactions, our design strategy aimed to increase the flexibility of E392 and enlarge the aromatic moiety of Y392 to enable both residues to bind simultaneously and strongly to one of the many basic residues at the floor of the receptor binding pocket along with additional modifications.

2.2. Peptide Stapling to Induce Helicity: Linker Positioning and Length

The most common approach for constraining peptides in an α‐helical conformation is by selective covalent stapling between amino acid side chains and is a rich and diverse field of research employing a wide variety of cyclisation chemistries [37]. We decided to explore stapling over two helical turns employing a copper‐catalyzed azide alkyne cycloaddition (CuAAC) (i, i+7)‐stapling strategy between two azido modified amino acid sidechains and a bis‐alkyne linker (1,3‐diethynylbenzene, DEB) as reported previously (Scheme 1 and Table 1) [29]. This particular stapling strategy has been shown by CD and NMR/MD to induce a high degree of α‐helicity and furthermore allows for functionalization on the linker if so desired by employing the complementary linker 1,3,5‐triethynylbenzene [29]. SPPS of linear bis‐azido functionalized peptides and following CuAAC stapling with DEB was uneventful and provided ample material for pharmacological characterization after RP‐HPLC purification to > 95% purity.

SCHEME 1.

Synthesis of stapled peptides. (A) and (B) CuACC (i, i+7)‐stapling of linear bis‐azido sidechain functionalized peptides 1–7 and 8 with DEB to give stapled peptides 1–7 and 8, respectively. (C) CuAAC (i, i+4)‐stapling of linear azido‐alkyne functionalized peptide 9 to give stapled peptide 9. The complete peptide sequences are shown in Table 1. CuAAC stapling was performed according to the published procedure [29] and was not optimized. The crude stapled peptides 1–9 were RP‐HPLC purified to obtain sufficient material for pharmacological characterization (> 95% purity). The numbers 380–394 denote the assigned residue numbers in the Gα_s protein.

TABLE 1.

Stapled peptides 1–9 for pharmacological characterization. The native C‐terminus (Gα_sCT₁₅) was also synthesized. The numbers 380–394 denote the assigned residue numbers in the Gα_s protein. Red residues indicate stapling positions. Blue residues indicate amino acid substitutions to the native Gα_sCT₁₅ peptide sequence to optimize ligand‐receptor interaction. Non‐canonical amino acids have been denoted B1–B7.

To mimic the helical conformation of Gα_sCT₁₅ in the crystal structures of the β₂AR‐G_s (Figure 1A) or β₂AR‐T4L‐G_sCT‐CC (Figure 1B) complexes, we sought to identify the optimal positions for stapling by computational methods. From the two available binding modes of Gα_sCT with the receptor obtained from x‐ray structure analysis, the interaction in β₂AR‐G_s(empty) seemed the more generalizable one as it has been reported for all Family A receptor‐G_s complex structures. To this end, we ranked the interactions of the α5 C‐terminus according to their stability in context of the receptor, obtained from previously performed MD simulations of these structures [33] to evaluate the potential impact of individual interactions for stability (Figure 1C). As potential staple positions, the first two (380+381) and last four (391–394) residues of GαsCT₁₅ were excluded to allow for some flexibility of the N‐terminal and C‐terminal parts. In the bound state, the α5 C‐terminus forms a reverse turn that caps the end of the helix satisfying the backbone hydrogen bonding potential of the ultimate helix turn. Thus, two stapling positions were considered, namely I383‐Q390 and I382‐R389, both principally allowing the key interactions of Y391 and E392 with the receptor. This design places the staple either laterally (I383‐Q390) or at the opposite side of these two residues (I382‐R389) (Figure 2A).

FIGURE 2.

Peptide cyclization to induce helicity. (A) Upper: Linker position at I383‐Q390 (green) or I382‐R389 (blue) in the Gα_sCT as part of the β₂AR‐G_s(empty) state (left) or β₂AR‐T4L‐G_sCT‐CC complex (right) shown from two different views, respectively. Lower: Visualization of potential attachment sites for the linker on the peptide sequence. Sites are colored according to their interaction with the receptor in the crystal structures of β₂AR‐G_s(empty) (light blue) or β₂Ar‐T4L‐G_sCT‐CC (violet). (B) Visualization of the bimane assay concept reporting receptor conformation: C265^6.27 in the lower TM6 of β₂AR is labeled with the highly environmentally sensitive bimane fluorophore (unconjugated λ _ex = 395 nm, λ _em = 480–500 nm), which is pointing toward the receptor core in an inactive receptor conformation. Upon receptor activation, TM6 rotates and moves outward, exposing the bimane fluorophore toward a more hydrophilic environment as show in (C) where increasing concentrations of the agonist isoproterenol (ISO) results in a bathochromic shift of λ _max along with a decrease of the fluorescence intensity of the C265^6.27 bimane [38]. The nanobody 80 (Nb80) can potentiate a submaximal ISO response (10 µM), whereas it has minimal endogenous effect when tested alone at 1 µM, indicating its ability to recognize and stabilize an active conformation of the β₂AR as a G_s protein surrogate as shown previously [27]. The Gα_sCT₁₅ linear peptide has no effect in presence of 10 µM ISO, whereas (i, i+7)‐stapling (and substitution of the oxidation prone methionine residue to the isostere norleucine) potentiates the stabilization of an active conformation induced by 10 µM ISO. Bimane fluorescence plots shown are representative of several individual repeats (see Table S9). Nb80 structure extracted from PDB ID: 3P0G [27].

In the β₂AR‐G_s(empty) binding mode, the I382‐R389 staple points toward the solvent, potentially allowing simultaneous hydrophobic interactions of the staple with TM5 and TM6 (Figure 2A). In preference of exploiting the potential hydrophobic interaction of the I382‐R389 position, we modeled two stapled peptides differing only in the length of the staple by two methylene units by introducing a staple using either azidohomoalanine (AHA) or azidoornithine (AzO, Table 1). Visual inspection suggested that the longer AzO‐based staple would be able to stabilize the α‐helical conformation, while preserving the internal flexibility of the peptide observed in MD simulations of the respective structures [33, 39]. By contrast, the shorter AHA‐based staple could potentially cause undesirable bending of the helical peptide or at least introduce more tension on its backbone and possibly remove the flexibility of the central helix part needed to accommodate the binding crevice. Consequently, we synthesized peptide 1 using the AzO modification. In addition to introducing the staple, the Gα_sCT₁₅ M386 residue was replaced by a norleucine (Nle), which is commonly used as a methionine mimic to circumvent side reactions from oxidation.

To assess the conformational state of the β₂AR stabilized by the peptide, we labeled C265^6.27 in the lower TM6 of β₂AR with the highly environmentally sensitive bimane fluorophore (unconjugated λ _ex = 395 nm, λ _em = 480–500 nm), which has previously been reported to change spectral properties depending on the conformational state of the β₂AR (Figure 2B) [38]. Shifting the equilibrium toward the active state by addition of increasing isoproterenol (ISO) agonist concentrations causes TM6 to rotate and move outward, toward a more hydrophilic environment, which resulted in a bathochromic shift of λ _max and a decrease in fluorescence intensity of the C265^6.27 bimane (Figure 2C). As previously observed [27], the G protein mimetic Nb80 potentiated the fluorescence shift by ISO (Figure 2C). Unlike the linear Gα_sCT₁₅ peptide, which at 20 µM did not potentiate the response of 10 µM ISO toward an active conformational state, the stapled peptide 1 potentiated the ISO response and had no effect when tested in the absence of ISO (Figure 2C). These data indicated that stapling of the Gα_sCT₁₅ peptide in an extended helical conformation is indeed beneficial for stabilization of an active‐like conformation in presence of an orthosteric agonist.

2.3. Peptide Residue Engineering to Increase the Stabilization of Receptor Active Conformations

To further optimize the binding properties of the peptide, we modified the two residues Y391 and E392 that interact with R131^3.50 of the preserved E(D)RY motif at the floor of the intracellular binding crevice in the two respective crystal structures (Figure 1A,B). Notably, at the intracellular binding pocket, a high number of basic residues (Figure 1D,E) are solvent exposed and thus available for energetically favorable interactions with Y391 or E392 or other acidic or aromatic residues from the G protein. To strengthen the anticipated interactions with basic residues from the intracellular binding pocket, two peptides were designed substituting Y391 with either 1‐naphthylalanine (Nal, peptide 2) or 3,4‐dimethoxy‐phenylalanine (dmPhe, peptide 5). Due to an extended aromatic system (Nal) or higher electron density (dmPhe), the two modifications could potentially strengthen binding of the peptide e.g. via cation–π interaction. In addition, both modifications were envisioned to increase the bulkiness at position Y391 potentially shifting TM6 in the active receptor conformation further outward. Also, we explored extension of the carboxylic group of the E392 side chain by one methylene unit by introduction of homoglutamic acid (hGlu, peptides 3 and 4), to enable ionic bonding or polar interactions with additional basic or polar residues (e.g. T66^2.37, R131^3.50, Q229^5.68, K270^6.32, R328^7.55, S329^8.47, and R333^8.51) of the binding pocket. In the two crystal structures, it is evident that either Y391 or E392 can interact with R131^3.50 at a time. We further envisioned that extension of the E392 side chain could facilitate simultaneous interaction of E392 and Y391 with R131^3.50 in the receptor.

When tested in the bimane assay, peptide 2 with the Y391Nal substitution substantially potentiated the 10 µM ISO response (Figure 3) compared to peptide 1 (Figure 2C), whereas peptide 5 with the Y391dmPhe substitution did not improve stabilization of the active conformation (Figure S1). Substituting E392 with side chain elongated hGlu (peptide 3) did not potentiate the ISO response beyond that of peptide 1. However, combining the Nal and hGlu substitutions in peptide 4 resulted in further potentiation of the ISO 10 µM bimane response, as compared to peptide 2. Importantly, the linear counterpart of peptide 2 lost its effect entirely indicating that both stapling and introduction of Nal are critical for the stabilization of an active conformation (Figure 3). Attempts to further stabilize the active receptor conformation by replacing additional residues in the peptide observed to interact in the β₂AR‐G_s(empty) structure were not successful. Substituting R380 with a Tyr in peptide 2 and peptide 4 yielding peptides 6 and 7, respectively, to potentially allow for extended interactions of the peptide with positively charged residues at ICL1 and TM5 and for π–cation–π stacking/stabilization with other bulky residues in this region (outlined in Figure S2) did not improve stabilization of the active conformation (Figure S1). The lack of effect by these substitutions provided the first indication that the peptides bind in a manner different from the pose in the β₂AR‐G_s(empty) structure.

FIGURE 3.

Stabilizing an active receptor conformation by peptide amino acid substitution. Substitution of residues Y391 and E392 were predicted to improve interaction of the peptide with the active receptor conformation. Substitution Y391Nal (peptide 2) improved the ability of the stapled peptide to stabilize an active receptor conformation induced by 10 µM ISO, whereas substitution E392hGlu (peptide 3) did not improve potentiation alone, but further stabilized an active‐like conformation when combined with the Nal substitution (peptide 4). Stapling was required for activity as the linear precursor of peptide 2 was not able to potentiate the ISO response. The staple position was also essential as moving the staple position from I383‐Q390 to I382‐R389 resulted in an inactive peptide (peptide 8). Bimane fluorescence plots shown are representative of several individual repeats (see Table S9).

Given the substantial improvement in the ability of the stapled peptide to stabilize an active conformation by the Y391Nal substitution, we also evaluated the importance of the stapling position between I382 and R389 by synthesizing peptide 8, an analog of peptide 2 in which the stapling position is shifted by one position toward residues I383 and Q390. However, peptide 8 completely lost potentiating effects compared to peptide 2, corroborating the initially selected I382‐R389 staple position that we favored to promote a hydrophobic interaction of the staple with TM5 and TM6 (Figure 3). Interestingly, employing the Y391Nal with a previously reported shorter (i, i+4) staple between M386 and Q390 [28] resulted in peptide 9 with an appreciable but lower potentiation of ISO in the bimane assay compared to peptide 2. The potentiation was also lost in the unstapled linear counterpart to peptide 9 (compare Figures S1 and 3). Interestingly, Ballet and coworkers have reported the synthesis of peptides with an (i, i+4)‐staple between positions 378 and 382 that were able to stabilize active‐like β₂AR conformations. However, unlike in the present study, neither incorporation of Nal at position 391 nor hGlu at position E392 improved the ability to stabilize the active conformation of β2AR [23]. Taken together, these data indicate that the combination of stapling of Gα_sCT₁₅ to conserve the α‐helical conformation at a specific position and strengthening of key interactions are required to efficiently stabilize an active‐like conformation of β₂AR.

2.4. In Vitro Characterization of the Peptidomimetics

The two most active peptides were characterized further in in vitro assays and compared to Nb80, a well‐characterized nanobody used as a G_s protein surrogate for stabilization of the β₂AR in an active conformation [27]. In absence of agonist, both peptides 2 and 4, as well as Nb80 showed only minor responses in the bimane assay suggesting they were unable to stabilize an active receptor conformation alone (Figure 4A–C). In presence of agonist (10 µM ISO) both peptides 2 and 4 showed saturable responses when tested up to 50 µM and the maximal shift was more efficacious to that obtained by a saturating concentration of Nb80 (3.2 µM) (Figure 4D–F). Peptides 2 and 4 at a concentration of 10 µM had a comparable shift to Nb80 tested at 1 µM. As observed when tested at a single concentration, peptide 4 appeared slightly more efficacious in bathochromic shifting of λ _max and decreasing fluorescence intensity compared to peptide 2 and Nb80.

FIGURE 4.

In vitro characterization of peptides 2 and 4 and comparison with G protein surrogate Nb80. In absence of ISO, Nb80 had marginal effects (A), whereas peptides 2 (B) and 4 (C) were unable to stabilize an active receptor conformation at the concentrations tested (testing at peptide concentrations higher than 50 µM was not possible due to limited peptide solubility). In presence of agonist (ISO 10 µM), Nb80 (D) and peptides 2 (E) and 4 (F) were able to stabilize an active‐like receptor conformation concentration‐dependently as indicated by the bimane assay. In a cAMP assay on membranes overexpressing the β₂AR, binding of Nb80 (G) and peptides 2 (H) and 4 (I) translated to a blockade of ISO‐induced cAMP formation. Notably, peptide 2 right‐shifted the ISO CRC along with decreasing E _max, whereas a decrease of E_max only was observed for Nb80 and peptide 4. cAMP data represents mean ± SEM from n = 3–6 experiments carried out in duplicate. The effect of all peptides (1–9, and linear 2+9) on cAMP ISO concentration response curves are given in Table S8. Bimane fluorescence plots shown are representative of several individual repeats (see Table S9).

A peptide able to bind in the same intracellular receptor crevice as the G protein would be expected to block G protein binding and activation. Inhibition of ISO‐induced cAMP formation was previously reported for Nb80 following its intracellular expression as Nb80 does not readily cross the cell membrane [40]. To enable comparison with Nb80 in a concentration dependent manner, peptides 2 and 4 were tested for inhibition of ISO‐induced cAMP formation in β₂AR membrane preparations where molecules have access to both extra‐ and intracellular surfaces of the receptor as described previously [26, 28]. In general, there was a correlation between the peptides’ ability to potentiate the bimane response by ISO and their ability to block ISO‐induced cAMP formation in the presence of Gα_s (Table S8). Peptides 2 and 4 (tested at 10 and 50 µM), and Nb80 (tested at 1 and 10 µM) lowered the E _max of ISO dose‐dependently (Figure 4G–I). Nb80 and peptide 4 did not shift the EC₅₀ of ISO but only decreased E _max, while peptide 2 right‐shifted the EC₅₀ and decreased E _max. Of note peptides 2 and 4 did not show any effect on ISO‐induced cAMP formation on cultured HEK293 cells expressing the β₂AR (Figure S3), indicating that the peptides do not readily cross the cell membrane and would likely require modification to enable membrane permeation, for example, with cell‐penetrating motifs such as Pep‐1, TAT or polyarginines [41].

To explore peptides 2 and 4 modulation of related G_s‐coupled receptors, the peptides were tested for activity in similar membrane preparations containing the β₁AR or dopamine D1 receptor (D₁R). While peptides 2 and 4 were also potent at the β₁AR they were less efficient for inhibition of D₁R‐induced cAMP formation (Figure S4, Table S10). Interestingly, revisiting peptides 6 and 7 containing a R380Y substitution, which did not improve β₂AR bimane assay shifts (Figure S1), were more active for blocking D₁R‐induced cAMP formation than 2 and 4 (Figure S4, Table S10). Further studies are needed to explore the structural basis for these observations but highlights the potential for development of receptor selective peptidomimetics. Although of high importance to inform on selectivity of the peptides toward G_s‐coupled receptors over G_i‐ and G_q‐coupled receptors, it was not possible to develop membrane‐based assays for G_i‐ and G_q‐coupled receptors.

2.5. Molecular Dynamics Simulation to Assess the Binding Mode of the Best Peptide

Having identified peptide 4 as a potent Gα_s protein mimetic for stabilizing the β₂AR in an active‐like conformation, we simulated spontaneous binding of the peptide to the receptor using all‐atomistic MD simulations to find suitable binding modes. The x‐ray structure of the β₂AR [PDB ID 3SN6] [31] bound to the high affinity agonist BI‐167107 was used for simulations. As in a recent analysis of association with the inactive G_s [34], we started our simulations with the D130^3.49–R131^3.50 ionic interaction preserved in the receptor, to facilitate intracellular binding of peptide 4. The forcefield parameters for the peptidomimetic, which includes non‐standard amino acids and components, were derived from extensive quantum mechanics calculations as described in the Methods section. Peptide 4 was initially placed ca. 30 Å from the intracellular receptor binding crevice to avoid imposing intramolecular contacts while allowing for free fluctuation and recognition (Figure S5A). As a first measure for structural integrity of peptide 4, in 20 replicates of 2 µs MD simulations started from this configuration, we analyzed the radius of gyration and RMSF before binding to the receptor (Figure 5A). Only the far C‐terminus of peptide 4 (L393 and L394) remained highly flexible. This is in agreement with MD simulations of the peptide bound to β₂AR‐G_s(empty) and β₂AR‐T4L‐G_sCT‐CC and confirms that the staple preserved the helicity of the peptide both for the stapled region and the flanking parts.

FIGURE 5.

Comparison of binding of Gα C‐terminal‐derived peptides to the active β₂AR. (A) Structural dynamics of the peptides obtained from MD simulations of linker‐stabilized peptide 4 unbound (yellow) or bound (orange) to the receptor, when compared to Gα_sCT unbound/in water (gray), bound to β₂AR‐T4L‐G_sCT‐CC state (violet) or β₂AR‐G_s(empty) (petrol). Left: RG (radius of gyration) and right: RMSF (root‐mean‐square‐fluctuation). (B) (i) Structural overlay of β₂AR conformations reveal different TM6 outward tilts in the inactive receptor (PDB ID: 2RH1, in gray) and the active receptor‐nanobody (in red, overlapping with violet and blue), β₂AR‐T4L‐G_sCT‐CC peptide (violet), β₂AR‐G_s (blue), β₂AR‐G_sGDP complexes (neon green) [36] and our peptide 4 complex (orange). (ii) Comparison of peptide 4 position obtained from MD simulations (orange, one representative frame from MetaDynamics walker3, cluster 265) to G_sCT in β₂AR‐G_s(empty) (blue) or β₂AR‐T4L‐G_sCT‐CC (violet) with the receptor shown as green cartoon. (C) Key interactions of E392hGlu, and Y391Nal and the hydrophobic patch of peptide 4 with the receptor in a representative pose, stabilizing an active conformation (figure from one representative frame from MetaDynamics walker3, cluster 265).

In our simulations, we observed a total of five different binding events of peptide 4 to the intracellular receptor crevice, during which the peptide comes in contact with the various intracellular loops ICL1‐3 and H8. In one trajectory (Figure S5B), peptide 4 binds deeply into the intracellular crevice of the receptor, which induces a rotameric switch of R131^3.50 that eventually breaks the D130^3.49–R131^3.50 ionic interaction. The disruption of this interaction is consistent with previous structural and biophysical analyses of rhodopsin activation [42] and is also observed in GDP bound and nucleotide‐free complex structures of the β₂AR with G_s [27, 34]. As anticipated, in the receptor bound mode, the flexibility of the far C‐terminus of peptide 4 was lost due to stabilizing interactions with the receptor (Figure 5A). To confirm this binding pose of peptide 4 as its lowest energy binding mode, we used metadynamics enhanced sampling that applies a history‐dependent bias to encourage the system to explore the free‐energy landscape (Figure S5A) [43]. Among other more transient binding modes visited during our metadynamics enhanced sampling, we found a low‐energy state in which the peptide is tightly bound to the intracellular crevice of the receptor (termed “β₂AR‐peptide 4 complex”), as in our initial MD simulation. In agreement with the in vitro pharmacology (vide supra), the receptor is stabilized in an active conformation in the MD simulations (Figure 5B,i). In the β₂AR‐peptide 4 complex, the TM6 is tilted more outward as in the crystal structure of the β₂AR in complex with Nb80, in agreement with the larger bathochromic shift of λ _max and decreased fluorescence intensity observed in the bimane assay with peptide 4 (Figures 3, 4F, and 5B,i). The outward tilt and kink of TM6 is stabilized by multiple interactions: The three residues Q390, Y391Nal and L394 of the C‐terminal half of peptide 4 form a scaffold that is locked in between the hydrophobic patch of TM6 and the basic residues of helix 8 (Figure 5C, Q390 is not shown). Y391Nal forms cation–π interactions with basic residues R328^7.55 and R333^8.51 in helix 8, but not with R131^3.50. Accordingly, the key binding mode observed in the nucleotide free complex is not observed in our simulations, where Y391Nal contacts in addition P330^8.48, I334^8.52, and Q337^8.55 (Figure S5B). An extended hydrophobic patch is formed between L388, L393, and L394 of peptide 4 with the hydrophobic patch of TM5 and TM6. Moreover, cation–π interactions are observed between the staple and the N‐terminal part of ICL3, which was modeled as an extension of TM5. Notably, the observed conformation would sterically be possible even with the missing 25 residues of the ICL3 that could form additional stabilizing interactions with the receptor.

In the β₂AR‐peptide 4 complex, the C‐terminus is oriented towards TM5 and TM6 (Figure 5B (ii),C), while the N‐terminus is rotated to a similar extent as the respective peptide in the β₂AR‐T4L‐G_sCT‐CC structure. A similar binding mode was observed in a recently proposed GDP‐bound β₂AR‐G_s intermediate that precedes the nucleotide‐free state [34]. As in this β₂AR‐G_s(GDP) intermediate, we observe a kink in TM6 at G280^6.42 from a glycine‐rich motif at the center of the helix in the β₂AR‐peptide 4 complex, resulting in a local tilt of 74°, which is larger than the local tilt angle of 35° measured for the β₂AR‐G_s(empty) state (PDB ID: 3SN6) (Figure S5C). In both the β₂AR‐G_s(GDP) and the β₂AR‐peptide 4 complex, T281^6.43 rotates into the center of the TM bundles to stabilize the TM6 kink by forming interactions with N322^7.49 (Figure S5B). Reminiscent of the binding mode in β₂AR‐G_s(GDP) and the β₂AR‐T4L‐G_sCT‐CC structure and in agreement with our design strategy, specific interactions of the acidic side chain moiety E392hGlu with R131^3.50 of the receptor are established (Figures 5C and S5B). The elongated E392hGlu side chain of the peptide points inward along the z‐axis of the receptor, toward R131^3.50, where it interacts with R328^7.55 and P330^8.48 of H8 and T274^6.36 and I278^6.40 of TM6 (Figure S5B). Finally, the C‐terminus of L394 stably interacts with K267^6.29 and K270^6.32 from the TM6 lysine–ladder (Figure S5B), that together with the E392‐R131^3.50 interaction play an important role in the association mechanism of GDP‐bound G_s protein with the β₂AR [34].

Supported by our MD simulations, we speculate that the peptidomimetics capable of stabilizing an active‐like β₂AR conformation in vitro, stabilize an otherwise transient, intermediate, active‐like conformational state preceding that of the β₂AR‐G_s(empty). The peptidomimetics also blocking cAMP formation appear promising as strong stabilizers and complimentary tools of hard‐to‐capture conformational states in non‐static studies of GPCR activation [44, 45].

3. Conclusion

Herein, we demonstrate that short 15‐mer peptidomimetics of the C‐terminal α5 helix of the Gα subunit in the G_s protein can be designed to effectively stabilize an active receptor conformation and block intracellular signaling of the β₂AR receptor. To this end, we applied a computational design strategy based on pre‐existing x‐ray crystallographic data. The rational design strategy led to the introduction of an (i, i+7)‐staple between positions I382 and R389 and sequence mutations to the C‐terminal end of the Gα_sCT₁₅. We found that the positioning and length of the staple was essential to achieve favorable interactions with the β₂AR. However, stapling alone only resulted in a small active conformation‐stabilizing effect. Computationally guided optimization resulted in the E392hGlu substitution to improve binding to R131^3.50 and other basic residues, and the Y391Nal substitution to strengthen cation–π interactions with R131^3.50 resulting in stapled peptide 4 that appear to stabilize an active‐like β₂AR conformation similar to that induced by G_s and Nb80 when monitoring TM6 movement. Further in vitro experiments support that peptide 4 binds to the intracellular crevice to block ISO‐induced cAMP production. MD simulations of peptide 4 with the β₂AR suggest a binding mode closer to the intermediate conformation seen in the β₂AR‐T4L‐G_sCT‐CC complex with E392hGlu interacting with R131^3.50, while including components from the β₂AR‐G_s(empty) complex and novel features such as a TM6 kink, the latter also captured in the receptor bimane conformational change assay. We further hypothesize that the additional bulkiness of Y391Nal aids to stabilize an active‐like β₂AR conformation with TM6 pushed outward.

In conclusion, we have shown that a computationally driven peptide design approach based on structural information could identify a potent nano G_s protein mimic that blocks the receptor from interacting with intracellular binding partners while stabilizing an active‐like conformation. The binding mode of the G_s protein mimic may represent another intermediate state preceding that of β₂AR‐G_s(empty), which can otherwise be hard to capture in an in vitro, non‐static context. We envision that this generalizable approach will be enable further structural and functional exploration of other GPCRs and possibly opens gateways toward novel therapeutics.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: Detailed material and methods description is in Supporting Information. The authors have cited additional references within the Supporting Information [28, 29, 31, 38, 46–79].

Data Availability Statement

Models, topologies, molecular dynamics input files, metadynamics input files, and other relevant data are available on Zenodo: https://doi.org/10.5281/zenodo.18085482. Selected molecular dynamics runs (see Supporting Information Section 2.3) can be accessed here: https://proteinformatics.uni‐leipzig.de/mdsrv.html?load = file://base/stapled_MWT/session_MWT.js.