Nanobody-Mediated Dualsteric Engagement of the Angiotensin Receptor Broadens Biased Ligand Pharmacology

Abstract

Dualsteric G protein-coupled receptor (GPCR) ligands are a class of bitopic ligands that consist of an orthosteric pharmacophore, which binds to the pocket occupied by the receptor’s endogenous agonist, and an allosteric pharmacophore, which binds to a distinct site. These ligands have the potential to display characteristics of both orthosteric and allosteric ligands. To explore the signaling profiles that dualsteric ligands of the angiotensin II type 1 receptor (AT1R) can access, we ligated a 6e epitope tag-specific nanobody (single-domain antibody fragment) to angiotensin II (AngII) and analogs that show preferential allosteric coupling to Gq (TRV055, TRV056) or β-arrestin (TRV027). While the nanobody itself acts as a probe-specific neutral or negative allosteric ligand of N-terminally 6e-tagged AT1R, nanobody conjugation to orthosteric ligands had varying effects on Gq dissociation and β-arrestin plasma membrane recruitment. The potency of certain AngII analogs was enhanced up to 100-fold, and some conjugates behaved as partial agonists, with up to a 5-fold decrease in maximal efficacy. Nanobody conjugation also biased the signaling of TRV055 and TRV056 toward Gq, suggesting that Gq bias at AT1R can be modulated through molecular mechanisms distinct from those previously elucidated. Both competition radioligand binding experiments and functional assays demonstrated that orthosteric antagonists (angiotensin receptor blockers) act as non-competitive inhibitors of all these nanobody-peptide conjugates. This proof-of-principle study illustrates the array of pharmacological patterns that can be achieved by incorporating neutral or negative allosteric pharmacophores into dualsteric ligands. Nanobodies directed toward linear epitopes could provide a rich source of allosteric reagents for this purpose.

SIGNIFICANCE STATEMENT

Here we engineer bitopic (dualsteric) ligands for epitope-tagged angiotensin II type 1 receptor by conjugating angiotensin II or its biased analogs to an epitope-specific nanobody (antibody fragment). Our data demonstrate that nanobody-mediated interactions with the receptor N-terminus endow angiotensin analogs with properties of allosteric modulators and provide a novel mechanism to increase the potency, modulate the maximal effect, or alter the bias of ligands.

Introduction

G protein-coupled receptor (GPCR) ligands can be broadly classified as either “orthosteric” ligands, which bind to the same pocket as endogenous agonists, or “allosteric” ligands, which bind myriad other sites (Thal et al., 2018). While the overwhelming majority of GPCR-targeted drugs are orthosteric ligands, allosteric ligands can display distinct and complementary attributes. In particular, allosteric ligands often exhibit greater selectivity for certain receptor subtypes, and they can tune the efficacy or signaling bias of orthosteric agonists.

A third, less explored class of GPCR ligands are dualsteric ligands, or bitopic ligands with pharmacophores binding to both the orthosteric site and an allosteric site (Antony et al., 2009; Lane et al., 2013). These have the potential to combine the desirable properties of orthosteric and allosteric ligands. However, the design of dualsteric compounds to target GPCRs using only low molecular weight building blocks (i.e., small molecules or short peptides) is challenging, largely due to the dearth of appropriate allosteric ligands. For most Family A (rhodopsin-like) GPCRs, the orthosteric site is the extracellular-facing pocket formed by the seven-transmembrane bundle, and there are limited other “druggable” extracellular pockets that can be easily engaged by small molecules. While our ability to rationally design orthosteric ligands based on structural studies is improving, allosteric binding sites are often not obvious from conventional analyses due to the complex conformational dynamics of these regions.

Antibodies (Abs) can alleviate some of the challenges associated with allosteric ligand discovery (Hutchings et al., 2017). Abs offer a large binding surface, which enables engagement with protein epitopes that are difficult to target with small molecules. Abs are highly specific, with the potential to recognize less conserved regions away from evolutionarily conserved orthosteric sites. Additionally, Abs can be generated against many targets, including GPCRs, through immunization and screening procedures. GPCR antibodies frequently exhibit excellent target affinity and specificity even when they act as neutral allosteric ligands that lack intrinsic agonist or antagonist activity.

Linking Abs to bioactive compounds offers a potentially favorable path toward dualsteric GPCR ligands. However, the production of Ab conjugates is laborious, requiring careful optimization of methods for conjugation chemistry that preserve the binding and signaling activity of their constituent pieces. Further, Ab production is costly, requiring production in specialized cell types or isolation from animal sera. Ab fragments, such as camelid-derived variable regions from heavy-chain-only antibodies (VHHs, or nanobodies, Nbs), provide a more tractable alternative (Sachdev et al., 2021). Nanobodies are single-domain antibody fragments (15 kDa) with affinities and selectivities comparable to conventional Abs (150 kDa), despite being approximately 10% as large (Muyldermans, 2013; Cheloha et al., 2020b). Their benefits include efficient bacterial expression, stability under adverse conditions, and the availability of straightforward labeling methods. Highly effective Nb-orthosteric ligand conjugates that target GPCRs have been published. A peptide ligand-Nb conjugate that targets parathyroid hormone receptor-1 showed enhanced signaling and specificity relative to the comparator peptide ligand (Cheloha et al., 2020a, 2021). In another example, semi-synthetic Nb-ligand conjugates demonstrated adjustable signaling properties and improved transcriptional outputs at neurokinin receptor-1 (Braga Emidio and Cheloha, 2023).

Here we use the angiotensin II type 1 receptor (AT1R) as a model to expand our understanding of how the pharmacology of nanobody-derived dualsteric ligands differs from parental orthosteric ligands. In addition to its therapeutic importance as the target of anti-hypertensive medications (angiotensin receptor blockers, ARBs), the AT1R is a particularly useful system because a spectrum of biased orthosteric ligands with well-defined structure-activity relationships exist. Specifically, analogs of the angiotensin II (AngII) octapeptide in which the C-terminal phenylalanine is deleted or replaced with a smaller side chain, such as TRV027, are able to activate β-arrestin-mediated but not Gq-mediated pathways (Wei et al., 2003; Violin et al., 2010). These “β-arrestin-biased ligands” cannot make key interactions at the base of the orthosteric binding site that trigger a cascade of conformational changes required for Gq activation (Wingler et al., 2020). Mutation of AngII Arg2 in “Gq-biased ligands,” such as TRV055 and TRV056, results in enhanced allosteric coupling to Gq, without the loss of β-arrestin coupling, as assessed through radioligand binding experiments (Strachan et al., 2014). Since the only reported extracellular AT1R nanobodies occupy the orthosteric site (McMahon et al., 2020), we employed an epitope tag-specific nanobody to provide an allosteric pharmacophore (Cheloha et al., 2020a). Using this engineered system, we demonstrate that the same allosteric nanobody can tune different combinations of potency, efficacy, and signaling bias when conjugated to a series of AngII analogs, and it can simultaneously confer properties of allosteric modulators upon these orthosteric ligands.

Materials and Methods

Materials

Unmodified AngII analogs were custom synthesized by GenScript (Piscataway, NJ). Cell culture media and reagents were obtained from Thermo Scientific (Waltham, MA) unless otherwise noted. Chemicals were purchased from MilliporeSigma (Burlington, MA) except as noted. Sources of other reagents are given below.

Cell Culture

Inducible Expi293 cells (Thermo Scientific, Waltham, MA) stably expressing the TetR repressor were maintained in Expi293 media containing 10 μg/ml blasticidin at 37°C/8% CO₂ in a humidified atmosphere with shaking at 110 rpm. Transfections were performed with Expifectamine according to the manufacturer’s instructions. Receptor expression was induced two days post-transfection with 4 μg/ml doxycycline (TCI, Portland, OR) and 5 mM sodium butyrate (Beantown Chemical, Hudson, NH), and signaling assays were performed the next day.

Cloning

To insert the 6e tag between the N-terminal FLAG tag and human AT1R, a 300-bp DNA fragment encoding the 6e tag (amino acid sequence: QADQEAKELARQIS) flanked by diglycine linkers and appropriate homology regions was obtained from Twist Bioscience (South San Francisco, CA), amplified with Vent polymerase (New England Biolabs, Ipswich, MA), and inserted into the plasmid pcDNA-Zeo-tetO-FLAG-AT1R (Wingler et al., 2019b) using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent, Santa Clara, CA). Correct insertion of the tag was confirmed by Sanger sequencing (Genewiz/Azenta Life Sciences, South Plainfield, NJ; RRID:SCR_003177).

Solid-Phase Peptide Synthesis

Peptides were synthesized on a Liberty Blue automatic synthesizer (CEM, Charlotte, NC, USA) via Fmoc-SPPS (9-fluorenylmethyloxycarbonyl-solid phase peptide synthesis; 0.05 mmol scale). G₃-AngII, G₃-TRV055, G₃-TRV056, and their derivatives were synthesized using Fmoc-Phe-Wang resin (ChemPep, Wellington, FL), while G₃-TRV027 and their derivatives were synthesized on a Fmoc-D-A-Wang resin (ChemPep, Wellington, FL). Deprotection of the Fmoc protecting group was performed using 10% piperidine/N,N-dimethylformamide (v/v). Couplings of the Fmoc-amino acids or building blocks [Fmoc-6-aminohexanoic acid (Fmoc-6-Ahx) or Fmoc-NH-(polyethylene glycol)₂-CH₂CH₂COOH (Fmoc-NH-PEG₂-CH₂CH₂COOH)] (10 equivalents) were carried out with N,N’-Diisopropylcarbodiimide/Oxyma Pure at 90°C (1: 2: 1 molar ratio of amino acid, N,N’-Diisopropylcarbodiimide, and oxyma). Upon completion of the peptide assembly, the resin was washed with dichloromethane. Cleavage from the resin and simultaneous removal of side-chain-protecting groups was achieved by treatment with 90% trifluoroacetic acid (TFA)/5% triisopropylsilane/5% H₂O at 25°C for 90 minutes. Following cleavage, the products were precipitated with cold diethyl ether and pelleted by centrifugation (3,000 RPM for 5 minutes). Then the peptides were lyophilized in 50% acetonitrile/0.1% TFA/H₂O. The crude products were purified by preparative reversed-phase high-performance liquid chromatography (RP-HPLC).

RP-HPLC and Liquid Chromatography–Mass Spectrometry

Peptide purification was performed using a preparative C18 column (Aeris PEPTIDE 5 µm XB-C18, LC Column 250 × 21.2 mm, AXIA Packed, Phenomenex, Torrance, CA) in a Shimadzu LC-20AR solvent delivery system (Kyoto, Japan). A gradient elution of 10–60% solvent B over 25 minutes was applied, with a flow rate of 10 ml/min. Solvents consisted of 0.1% TFA in H₂O (solvent A) and 0.1% TFA in acetonitrile (solvent B). The molecular weight of the fractions collected was evaluated on a Waters Xevo qTOF LC/MS (Milford, MA). Samples were resolved by RP-HPLC on a Hamilton PRP-h5 column (5 μM particle size, 300 Å pore size) (Reno, NV) and analyzed in positive ion mode. Data acquisition and processing were carried out using MassLynx software (Waters, Milford, MA; RRID:SCR_014271). Fractions containing the desired mass were subjected to RP-HPLC to analyze their purity prior to being lyophilized.

Nanobody Recombinant Expression

WK6 E. coli were heat shock transformed with a pHEN6 plasmid encoding Nb6e and cultured in Terrific Broth medium containing ampicillin (100 μg/ml). A preculture of the transformed bacteria was used to inoculate a full-size culture (2 L), which was shaken at 37°C until achieving an optical density at 600 nm between 0.6 and 0.8. Protein expression was induced with isopropyl-D-1-thiogalactopyranoside (IPTG, 1 mM) and the induced culture was shaken at 30°C overnight. Bacterial pellets were collected by centrifugation at 6,000 RPM (Avanti J Series centrifuge, Beckman Coulter, Brea, CA) for 20 minutes and resuspended in nickel nitrilotriacetic acid (NTA) wash buffer (tris buffered saline (50 mM Tris, 150 mM NaCl, pH 7.5) + 10 mM imidazole, pH 7.5) containing lysozyme and incubated on ice for 10 minutes. The cells were then lysed through three rounds of sonication. The lysate was then centrifuged at 16,000 RPM for 40 minutes to pellet the lysate. Nb6e was then purified from the supernatant by affinity chromatography. The supernatant passed through a fritted column containing nickel NTA beads (His Pur Ni-NTA Resin, Thermo Scientific, Waltham, MA) pre-equilibrated with nickel NTA wash buffer. The beads were then washed with nickel NTA wash buffer (3x). Next, the bound Nb6e was eluted with 10 ml of nickel NTA elution buffer (tris-buffered saline + 150 mM imidazole, pH 7.5). Nb6e was further purified by size-exclusion (HiLoad TM 16/600 Superdex 200 pg column, Cytiva Akta/Pure, Marlborough, MA) using an isocratic gradient of tris-buffered saline at 1 ml/min flow rate. Spin filtration columns (Amicon Ultra-15, regenerated cellulose, 10 kDa nominal molecular weight limit) were used to concentrate the fractions containing Nb6e. Protein concentration was quantified using a Nanodrop spectrometer (Thermo Scientific, Waltham, MA; RRID:SCR_020309) by ultraviolet spectroscopy measuring absorption at 280 nm.

Nanobody Conjugation via Sortagging

G₃-AngII, G₃-TRV027, G₃-TRV055, G₃-TRV056, and their derivatives were conjugated to Nb6e via sortase-A-mediated ligation (sortagging reactions). The reactions were carried out in sortase buffer (10 mM CaCl₂, 50 mM Tris, 150 mM NaCl, pH 7.5) containing the Nb6e bearing a sortase recognition motif (LPETGG) (200 μM), triglycine functionalized peptide (1000 μM), and Sortase-A 5M (20 μM) for 16 h at 12°C under agitation. Since the C-terminal His6 tag of Nb6e is lost upon peptide ligation, the reaction then was incubated with nickel NTA beads to capture Sortase-A 5M and unreacted Nb6e. Triglycine-peptides were removed from Nb6e conjugates using disposable desalting columns (Cytiva PD-10 Sephadex G-25M, Marlborough, MA). Spin filtration (Amicon Ultra 0.5 mL Centrifugal Filters 10 kDa nominal molecular weight limit, MilliporeSigma, Burlington, MA) was used to concentrate the fractions containing Nb6e conjugates. Liquid chromatography-mass spectrometry was used to confirm the molecular weight of the conjugates. Conjugate concentrations were measured by ultraviolet spectroscopy measuring absorption at 280 nm using a Nanodrop spectrometer (Thermo Scientific, Waltham, MA). We did not observe any adverse effects of peptide conjugation on nanobody stability or solubility.

Gq Activation Assays

Plasmids for the TRUPATH system were a gift from Bryan Roth (Addgene kit #1000000163). Inducible Expi293 cells were co-transfected with equal amounts of Gα_q-mutant Renilla luciferase (RRID:Addgene_140982), Gβ₃ (RRID:Addgene_140988), Gγ₉-mutant green fluorescent protein (RRID:Addgene_140991), and either pcDNA-Zeo-tetO-Flag-AT1R (“wild-type”) or pcDNA-Zeo-tetO-Flag-6e-AT1R (“6e-AT1R”). Immediately before beginning the assay, cells were harvested, resuspended in Hanks’ balanced salt solution (MilliporeSigma, Burlington, MA) + 20 mM HEPES pH 7.4 (assay buffer), and dispensed into opaque white 96-well plates. Cells were incubated with 1.3 μM Prolume Purple (methoxy e-Coelenterazine) (Nanolight Technologies, Prolume, LTD., Pinetop, AZ) for 5 minutes before the addition of a serial dilution of peptide or nanobody conjugate ligands in assay buffer containing 0.1% bovine serum albumin. Plates were read ten times on a Biotek Synergy Neo2 plate reader (Agilent, Santa Clara, CA; RRID:SCR_019765) set at 37°C with 400 and 510 nm emission filters, with an integration time of 1 second per well. For assays performed in antagonist mode, cells were pre-treated with DMSO or a dose-response of losartan for 15 minutes before the addition of Prolume Purple. Using data from the sixth read, the bioluminescence resonance energy transfer ratio was calculated from the ratio of the emission at 510 nm (mutant green fluorescent protein) to the emission at 400 nm (mutant Renilla luciferase).

β-Arrestin Recruitment Assays

The LgBit-CAAX and β-arrestin2-SmBit plasmids were a gift from Sudar Rajagopal (Duke University). Inducible Expi293 cells were co-transfected with a 1:4:5 ratio of LgBit-CAAX, β-arrestin2-SmBit, and either pcDNA-Zeo-tetO-Flag-AT1R (“wild-type”) or pcDNA-Zeo-tetO-Flag-6e-AT1R (“6e-AT1R”). Immediately before beginning the assay, cells were harvested, resuspended in Hanks’ balanced salt solution + 20 mM HEPES pH 7.4 (assay buffer) containing Coelenterazine-H (Research Products International, Mount Prospect, IL) (2.5 μM final concentration after ligand addition), and dispensed into opaque white 96-well plates. Basal luminescence was read three times on a Clariostar plate reader (BMG LabTech, Cary, NC). Cells were treated with a serial dilution of peptide or nanobody conjugate ligands in assay buffer containing 0.1% bovine serum albumin. Luminescence was read every minute for 50 minutes. Using data from the read 20 minutes post-stimulation, luminescence values from each well were normalized to the average of the three basal luminescence reads. For assays performed in antagonist mode, cells were pre-treated with DMSO, nanobody, or a dose-response of losartan for 15 minutes before the addition of agonists. DN1, a nanobody specific for mGluR2 (Scholler et al., 2017), was used as a negative control to test the effects of unconjugated Nb6e on parental peptides.

Radioligand Binding

Crude membranes were prepared from Inducible Expi293 cells as previously described (Strachan et al., 2014). Binding reactions were performed in a total volume of 200 μl in assay buffer consisting of 50 mM Tris pH 7.4, 150 mM NaCl, 12.5 mM MgCl₂, 0.2% bovine serum albumin, and the protease inhibitors leupeptin and benzamidine. For competition binding experiments, membranes were incubated with a half-log serial dilution of peptide or nanobody conjugate ligand and 1.2 nM [³H]-olmesartan (10 Ci/mmol, American Radiolabeled Chemicals, St. Louis, MO), with a single technical replicate per condition. To determine the effect of unconjugated Nb6e on AngII and TRV027 affinity, 1 μM Nb6e was also included in the reaction. For saturation binding experiments, membranes were incubated with or without 1 μM unconjugated Nb6e and a titration of [³H]-olmesartan in duplicate. The concentration of each [³H]-olmesartan dilution was determined empirically by scintillation counting. Candesartan (10 μM) was used to determine non-specific binding at each radioligand concentration. After a 90-minute incubation at room temperature, membranes were collected on GF/B filters using a Brandel 96-well harvester (Gaithersburg, MD) and rapidly washed three times with cold 50 mM Tris, pH 7.4. Bound radioactivity was measured using a Packard TriCarb 2100TR scintillation counter (Meriden, CT).

Data Analysis

The experiments reported in this manuscript were exploratory in nature and not designed to test a prespecified null hypothesis. In all experiments, data analysis was based upon the biologic replicates. For Gq and β-arrestin activation assays, responses were normalized to those of cells treated with vehicle (0%) and 1 μM AngII (100%) on the same microtiter plate. A 3-parameter dose-response curve was fit to the normalized data from 3 independent experiments in GraphPad Prism v. 10.0.0 (San Diego, CA; RRID:SCR_002798). Non-overlapping 95% confidence intervals (CIs) for EC₅₀ and maximal efficacy (E_max) values were considered to be statistically significant. For losartan inhibition experiments, Gaddum/Schild analysis of data obtained with parental and G3-modified peptides, as well as Nb6e-Ahx-AngII with wild-type AT1R, was performed in GraphPad Prism; the values of the Schild slope for losartan and the Hill slope of the agonist were not constrained. Data obtained with Nb6e conjugates in losartan inhibition experiments were analyzed using an operational model of allosterism (Leach et al., 2007) in GraphPad Prism, constraining the values of τ_B = 10⁻¹⁰⁰ (intrinsic efficacy of the “allosteric” antagonist), β = 0.01 (parameter describing the effect of the “allosteric” modulator on agonist efficacy), and E_max to the value obtained from the standard 3-parameter dose-response curve fit for that ligand in the absence of antagonist treatment (Supplemental Table 1).

For competition radioligand binding assays, a one-site competition binding model was fit to the data from each experiment in GraphPad Prism. The amount of bound radioligand was normalized to the top (100%) and bottom (0%) of the competition curve of the parental ligands (TRV055 was used for TRV056 and its derivatives due to the low affinity of TRV056). Inhibition constant (K_i) values were determined for peptides using the experimentally determined radioligand concentration in each experiment. The reported mean and 95% CIs of the LogK_i(-Nb6e)/LogK_i(+Nb6e) ratios were determined from three independent experiments.

For saturation radioligand binding assays, a one-site saturation binding model was fit to the data from each individual experiment in GraphPad Prism using the experimentally determined concentrations of [³H]-olmesartan. Reported dissociation constant (K_d) values represent the mean and 95% CIs determined from three independent experiments.

Results

Engineering Dualsteric AT1R Ligands

To explore how the addition of interactions with the AT1R N-terminus would affect the pharmacology of peptide ligands, we exploited the high-affinity interaction (K_d ∼10 nM) between Nb6e and the 6e peptide, a 14-amino acid tag derived from E2 ubiquitin-conjugating enzyme UBC6e (Ling et al., 2019; Cabalteja et al., 2022a). We have previously used this system to create dualsteric ligands for other GPCRs (Braga Emidio and Cheloha, 2023; Cabalteja et al., 2022a,b). We inserted the 6e tag between a FLAG tag and residue 2 of human AT1R (6e-AT1R) (Fig. 1A). To validate 6e-AT1R, we transiently transfected suspension HEK293 cells with the parental FLAG-AT1R construct (referred to as “wild-type AT1R”) or 6e-AT1R and with biosensors that allow us to assess transducer activation following agonist stimulation. Activation of heterotrimeric Gq protein was monitored using TRUPATH (Olsen et al., 2020), an optimized bioluminescence energy transfer assay that detects dissociation of Gα-mutant Renilla luciferase (engineered Renilla luciferase) and mutant green fluorescent protein-Gγ fusions (Gales et al., 2006; Sauliere et al., 2012). β-Arrestin recruitment to the plasma membrane was assessed using a bystander split NanoLuc assay, in which complementation of β-arrestin2-SmBiT and LgBiT-CAAX reconstitutes a functional luciferase enzyme (Dixon et al., 2016; Xu et al., 2022). Importantly, these two assays are both proximal measures of transducer activation that minimize signal amplification, do not require modification of the receptor, and can be carried out in the same cell line. We found that the 6e tag does not alter the potency of AngII in either Gq or β-arrestin activation (Fig. 1, B and C, Supplemental Table 1).

Fig. 1.

Design of dualsteric AT1R ligands. (A) Schematic of 6e-tagged AT1R construct. (B,C) Comparison of AngII-induced activation of (B) Gq dissociation and (C) β-arrestin2 recruitment to the plasma membrane by transiently transfected wild-type AT1R (FLAG-AT1R lacking the 6e epitope) or the 6e-AT1R construct shown in panel A. (D) Effect of a 15-minute pre-incubation with 1 μM control nanobody or unconjugated Nb6e on β-arrestin2 plasma membrane recruitment by the indicated ligands at 6e-AT1R. Curve fit parameters are provided in Supplemental Table 3. Data in panels B–D are normalized to the response induced by 1 μM AngII and represent the mean ± S.D. of three independent experiments each performed with two technical replicates. (E) Synthesis of Nb6e-AT1R ligand conjugates by sortase A (Srt A)-mediated ligation. AT1R ligands with the indicated pharmacological profiles and peptide sequences were modified at their N-termini with a triglycine motif (G₃) and one of the linkers shown. The N-terminal G₃ moiety enables ligands to be ligated to Nb6e with a C-terminal “LPETGG” tag, followed by a hexahistidine (H6) epitope tag that is removed upon ligation.

We then tested whether Nb6e is an allosteric modulator of 6e-AT1R. In radioligand binding assays, 1 μM Nb6e had no effect on the affinity of AngII and TRV056, and only a minimal effect on TRV027 and TRV055 affinity for 6e-AT1R (1.3- and 1.5-fold increase in K_i values, respectively) (Supplemental Table 2). In β-arrestin recruitment assays with 6e-AT1R, Nb6e did not alter AngII signaling substantially, but it increased the EC₅₀ of TRV027, TRV055, and TRV056 by approximately fivefold (Fig. 1D, Supplemental Table 3). A control nanobody that does not recognize AT1R had no effect in this assay. Thus unconjugated Nb6e behaves as probe-specific neutral or negative allosteric modulator of these orthosteric ligands.

The dualsteric AT1R ligands were produced by conjugating the peptides AngII, TRV027, TRV055, TRV056 to Nb6e via sortagging (Fig. 1E). In this method, the enzyme Sortase-A cleaves the peptide bond between the amino acid residues threonine and glycine in a specific recognition sequence (LPXTG motif). This cleavage generates a thioester-linked intermediate on the enzyme that undergoes a reaction with another fragment containing an N-terminal triglycine group, resulting in a covalent linkage between the two fragments. To allow for the conjugation of AT1R ligands to Nb6e, we expressed Nb6e with a sortase recognition motif (LPETGG) at its C-terminus, while AngII, TRV027, TRV055, and TRV056 were chemically synthesized with a triglycine motif at their respective N-termini (Supplemental Fig. 1, A and B). The initial conjugates tested contained a 6-carbon spacer amino acid (6-aminohexanoic acid, Ahx) between the triglycine moiety and the ligand. The formation of Nb-peptide conjugates was confirmed by mass spectrometry analysis (Supplemental Fig. 1C).

Nb6e Conjugation Differentially Modulates AT1R Ligand Potency and Efficacy

We compared the ability of the parental AT1R ligands, the unconjugated G₃-Ahx-modified peptides, and the Nb6e-Ahx-peptide conjugates to activate Gq and β-arrestin using the assays described above (Fig. 2, Supplemental Table 1). In all experiments, ligand-dependent responses were expressed as the percentage of that induced by 1 μM AngII for the same receptor construct on the same microtiter plate. The G₃-Ahx modification decreased the potency (EC₅₀) of AngII in Gq activation by approximately fivefold at both 6e-AT1R and wild-type AT1R (Fig. 2, A and B, Supplemental Table 1). Nb6e conjugation rescued this effect only for 6e-AT1R, so that the conjugate’s potency and efficacy (E_max) were indistinguishable from that of AngII. While the G₃-Ahx modification did not affect the activity of AngII in β-arrestin recruitment, Nb6e conjugation decreased ligand potency at wild-type AT1R (Fig. 2D). Here again, addition of the 6e tag to AT1R fully reversed this impairment (Fig. 2C). In both the Gq and β-arrestin assays, Nb6e-Ahx-AngII acted as a full agonist at 6e-AT1R (Fig. 2, A and C, Supplemental Table 1).

Fig. 2.

Functional activity of Nb6e-Ahx-AT1R ligand conjugates. (A–P) Activity of the parental AT1R ligands, the G₃-Ahx-modified ligands, and the Nb6e-Ahx-ligand conjugates in assays for Gq dissociation (left half of figure) and β-arrestin2 recruitment to the plasma membrane (right half of figure) in cells expressing either 6e-AT1R or wild-type AT1R as indicated. Data are normalized to the response induced by 1 μM AngII and represent the mean ± S.D. of three independent experiments each performed with two technical replicates. Curve fit parameters are provided in Supplemental Table 1.

Like the parental β-arrestin-biased ligand TRV027, G₃-Ahx-TRV027 and Nb6e-Ahx-TRV027 did not detectably activate Gq (Fig. 2, E and F). As reported in similar assays (Kawakami et al., 2022), TRV027 behaved a partial agonist of β-arrestin recruitment compared with AngII (Fig. 2, G and H, Supplemental Table 1). While the G₃-Ahx linker reduced TRV027 potency by over 2 orders of magnitude, the EC₅₀ of Nb6e-Ahx-TRV027 at 6e-AT1R was comparable to that of TRV027 (Fig. 2G, Supplemental Table 1). However, Nb6e conjugation decreased the efficacy of TRV027 by approximately 2-fold.

The Gq-biased ligands TRV055 and TRV056 showed the most dramatic changes upon conversion into dualsteric ligands. The potencies of Nb6e-Ahx-TRV055 and Nb6e-Ahx-TRV056 in Gq activation at 6e-AT1R were greatly enhanced, with EC₅₀ values approximately 100-fold lower than TRV055 and TRV056 (Fig. 2, I and M, Supplemental Table 1). A smaller (<10-fold) decrease in the EC₅₀ for β-arrestin recruitment was observed for these nanobody conjugates (Fig. 2, K and O, Supplemental Table 1). Interestingly, Nb6e-Ahx-TRV055 behaved as a full agonist relative to TRV055 (Fig. 2, I and K, Supplemental Table 1), but Nb6e-Ahx-TRV056 showed a reduction in efficacy versus TRV056 in both Gq and β-arrestin activation (Fig. 2, M and O, Supplemental Table 1).

The notable difference in the increased potency of Nb6e-Ahx-TRV055 and Nb6e-Ahx-TRV056 toward Gq versus β-arrestin prompted us to assess how the conjugation of Nb6e affected ligands’ signaling bias. We constructed bias plots for each pair of parental and Nb6e-conjugated ligands, plotting β-arrestin responses versus Gq responses at the same ligand concentrations. Bias plots enable the visual assessment of signaling bias between different ligands based on differences in the curve slope and shape (Gregory et al., 2010; Zheng et al., 2022). In contrast to Nb6e-Ahx-AngII and Nb6e-Ahx-TRV027, the responses of Nb6e-Ahx-TRV055 and Nb6e-Ahx-TRV056 were markedly shifted toward Gq when compared with TRV055 and TRV056, respectively (Fig. 3).

Fig. 3.

Bias plots of Nb6e-Ahx-AT1R ligand conjugates. (A–D) Using the data shown in Fig. 2, the normalized β-arrestin2 recruitment response is plotted versus the normalized Gq dissociation response for the same concentration of ligand. Data represent the mean ± S.D. of three independent experiments each performed with two technical replicates.

The Gq bias of the Nb6e conjugates was assessed quantitatively by calculating bias factors (β), defined as the logarithm of the ratio of the Nb6e conjugates’ and the parental ligands’ intrinsic relative activities at each pathway (Ehlert, 2008):

The bias factor of Nb6e-Ahx-TRV056 (β = 1.0, 95% CI: 0.6–1.4) indicates that, compared with TRV056, it promotes Gq dissociation 10 times more effectively than β-arrestin recruitment. Nb6e-Ahx-TRV055 displayed an even stronger Gq bias (β = 1.3, 95% CI: 1.0–1.7). These two ligands also showed a smaller bias toward Gq when compared with the endogenous agonist AngII (Supplemental Table 4). Nb6e-Ahx-AngII showed a smaller trend toward Gq bias (β = 0.4, 95% CI: 0.0–0.8), but its bias factor did not reach significance. Since TRV027 and Nb6e-Ahx-TRV027 did not generate a significant response in the Gq assay, a bias factor could not be determined. Interestingly, in contrast to the trend toward Gq bias observed when comparing the Gq-mediated production of inositol 1-phosphate to β-arrestin recruitment to AT1R (Rajagopal et al., 2011), TRV055 and TRV056 show a bias toward β-arrestin in our assays (Supplemental Table 4). This is likely a manifestation of “system bias” and demonstrates how cell type-specific differences in variables such as transducer expression and choice of readout can influence assessments of ligand bias (Onaran et al., 2014; Smith et al., 2018).

Dualsteric Nb6e-Ligand Conjugates Do Not Fully Occupy the Orthosteric Ligand-Binding Pocket

To better understand how Nb6e modulates the interactions of conjugated peptides with the orthosteric ligand-binding site, we tested the ability of Nb6e-ligand conjugates to block binding of the orthosteric small-molecule ARB [³H]-olmesartan (Fig. 4, A-D). Despite its full efficacy in signaling experiments (Fig. 2, A and C), at saturation, Nb6e-Ahx-AngII was only able to block approximately 20% of [³H]-olmesartan binding in competition radioligand binding experiments using crude cell membranes obtained from 6e-AT1R-overexpressing cells (Fig. 4A, Supplemental Table 1). However, the IC₅₀ value of the competition curve (7.2 nM, LogIC₅₀ = −8.1, 95% CI: −8.6−7.6) indicated a high affinity close to that of the Nb6e/6e tag interaction (∼10 nM) (Cabalteja et al., 2022a). This pattern of partial blockade was specific to the Nb6e conjugate, since the modified peptide G₃-Ahx-AngII acted as a competitive ligand with reduced affinity relative to AngII. Nb6e-Ahx-TRV055 and Nb6e-Ahx-TRV056 displayed similar patterns of partial competition with [³H]-olmesartan (Fig. 4, C and D, Supplemental Table 1).

Fig. 4.

Interaction of Nb6e-Ahx-AT1R ligand conjugates with the orthosteric ligand-binding site. (A–D) Competition of ligands with [³H]-olmesartan in cell membranes overexpressing 6e-AT1R. Data from each individual experiment are normalized to the top and bottom of the competition curve of the parental ligands. Data for in panel D were normalized to the TRV055 competition curve due to the low affinity of TRV056. Data represent the mean ± S.D. of three independent experiments performed with single technical replicates. Curve fit parameters are provided in Supplemental Table 1. (E) Model of competition between the orthosteric [³H]-olmesartan RL and dualsteric Nb-peptide conjugates. The cooperativity factor α describes the effect that nanobody binding to the allosteric site only has on radioligand binding. (F) Simulation of competition binding curves for dualsteric Nb-peptide conjugates with varying values of K_Nb+pep based on the model shown in panel E, eq. 2, the indicated values of α and K_Nb, and the experimentally determined values of K_RL and [RL] (see main text).

No decrease in [³H]-olmesartan binding was observed even at the highest concentrations of Nb6e-Ahx-TRV027 achievable in these experiments (Fig. 4B). The maximum concentration was two orders of magnitude above those required to stimulate β-arrestin recruitment detectably in cell-based assays (Fig. 2G). While G₃-Ahx-TRV027 blocked only a small fraction of radioligand binding at the highest dose tested (Fig. 4B), this is likely to be due to greatly reduced affinity compared with TRV027, in keeping with the modified peptide’s weak potency toward β-arrestin (Fig. 2G).

May et al. previously developed a model of how dualsteric ligands will modulate the binding of an orthosteric radioligand (RL) (Fig. 4E) (May et al., 2007). It can be applied to our allosteric Nb-orthosteric peptide (pep) conjugate as follows:

Here, K_RL represents the equilibrium dissociation constant of the radioligand in the absence of other bound ligands, K_Nb represents the equilibrium dissociation constant of the Nb-peptide ligand when it is bound only to the allosteric 6e tag through the nanobody, and K_Nb+pep represents the equilibrium dissociation constant of Nb-peptide when it is bound to both the allosteric and orthosteric sites, thus sterically occluding radioligand binding.

In this model, dualsteric ligands may exhibit a range of effects on radioligand binding that are determined both by the cooperativity factor α (here, the allosteric effect of nanobody binding on radioligand binding) and the relative values of K_Nb and K_Nb+pep (May et al., 2007). We anticipated that Nb6e would not allosterically affect binding of the radioligand (α = 1) since the N-terminus of the AT1R does not participate in the binding of small molecule ARBs (Zhang et al., 2015). Consistent with this assumption, a saturating concentration (1 μM) of unconjugated Nb6e to 6e-AT1R membranes did not significantly alter the affinity of [³H]-olmesartan in saturation binding experiments (K_d,−Nb6e = 500 pM, 95% CI: 400–600 pM; K_d,+Nb6e = 600 nM, 95% CI: 500–700 pM) (Supplemental Fig. 2). Since peptide ligands are ligated to the C-terminus of Nb6e, well removed from its complementarity determining region, it is also unlikely that conjugation affects K_Nb. This leaves K_Nb+pep, the dissociation constant of the two-site interaction, as the principal unknown.

Using these fixed values for α, K_Nb, and K_RL, and the known radioligand concentration, we simulated the binding of dualsteric ligands with varying values of K_Nb+pep using eq. 2 (Fig. 4F). Under conditions where K_Nb+pep = K_Nb, the dualsteric ligand would block only 23% of radioligand binding with an IC₅₀ = 8 nM. This pattern is strikingly close to the experimentally observed Nb6e-Ahx-AngII, Nb6e-Ahx-TRV055, and Nb6e-Ahx-TRV056 competition curves (Fig. 4, A, C, and D). If the two-site binding mode has lower affinity than the nanobody alone, K_Nb+pep becomes greater than K_Nb, and no competition with the radioligand would be observed (e.g., K_Nb+pep = 10K_Nb in Fig. 4F). This is consistent with the data for Nb6e-Ahx-TRV027 (Fig. 4B). The two-site binding mode would need to have substantially higher affinity than the nanobody alone (K_Nb+pep ≪ K_Nb) for Nb6e-peptide conjugates to quantitatively inhibit [³H]-olmesartan binding (Fig. 4F).

The observed pattern of radioligand binding could arise from direct allosteric effects of Nb6e binding (i.e., inducing conformational changes in the AT1R that alter peptide ligand binding) or from the relative spatial orientation of the two tethered moieties. As mentioned above, unconjugated Nb6e has minimal or no effects on the binding of parental ligands (Supplemental Table 2). These data indicate that allosteric effects cannot fully account for the difference in K_Nb and K_Nb+pep predicted by our binding model and competition binding data, suggesting that it is principally an emergent property due to nanobody-peptide conjugation.

Linker Length Does Not Govern the Pharmacological Profiles of Nb6e-Peptide Conjugates

The nature of the linkage between pharmacophores of dualsteric ligands can alter many aspects of their functional properties. Accordingly, we removed the Ahx linker of Nb6e-Ahx-AngII and Nb6e-Ahx-TRV027, leaving only the shorter linker “G₃” in which the peptide is directly conjugated to the sortase recognition motif, or we replaced the Ahx linker with a substantially longer (Ahx₂) or a more hydrophilic (PEG₂) linker (Fig. 1E). For the Nb6e-linker-AngII series, none of the variants significantly altered ligand potency in either Gq or β-arrestin activation, and all conjugates retained full agonist activity (Fig. 5, Supplemental Table 1). Like Nb6e-Ahx-AngII (Fig. 4A), these ligands blocked approximately 15–20% of [³H]-olmesartan binding in competition experiments (Fig. 5, Supplemental Table 1), suggesting that altering linker length did not increase or decrease the affinity of the two-site-binding interaction (Fig. 4, E–F).

Fig. 5.

Characterization of Nb6e-AngII conjugates with varying linker lengths. (A–C) Activity of the parental AT1R ligands, the G₃-linker-modified ligands, and the Nb6e-linker-AngII conjugates in assays for Gq dissociation (left panels) and β-arrestin2 recruitment to the plasma membrane (middle panels) in cells expressing 6e-AT1R. Data are normalized to the response induced by 1 μM AngII and represent the mean ± S.D. of three independent experiments each performed with two technical replicates. Competition of ligands with [³H]-olmesartan in cell membranes overexpressing 6e-AT1R is shown in the right panels. Data from each individual experiment are normalized to the top and bottom of the AngII competition curve. Data represent the mean ± S.D. of three independent experiments performed with single technical replicates. Curve fit parameters for all experiments are provided in Supplemental Table 1.

For the Nb6e-linker-TRV027 series, the G₃, PEG₂, and Ahx₂ variants had EC₅₀ values similar to Nb6e-Ahx-TRV027 in β-arrestin recruitment, but their efficacies were all slightly lower than the original conjugate (Fig. 6, Supplemental Table 1). Neither increasing nor decreasing the linker length enabled these ligands to compete with [³H]-olmesartan in competition binding assays. Together, these data show that within the range of linker lengths tested, the effects of Nb6e conjugation on orthosteric peptide binding and signaling efficacy are primarily an intrinsic property of the peptide structure.

Fig. 6.

Characterization of Nb6e-TRV027 conjugates with varying linker lengths. (A–C) Activity of the parental AT1R ligands, the G₃-linker-modified ligands, and the Nb6e-linker-TRV027 conjugates in β-arrestin2 recruitment to the plasma membrane (left panels) in cells expressing 6e-AT1R. Data are normalized to the response induced by 1 μM AngII and represent the mean ± S.D. of three independent experiments each performed with two technical replicates. Competition of ligands with [³H]-olmesartan in cell membranes overexpressing 6e-AT1R is shown in the right panels. Data from each individual experiment are normalized to the top and bottom of the TRV027 competition curve. Data represent the mean ± S.D. of three independent experiments performed with single technical replicates. Curve fit parameters for all experiments are provided in Supplemental Table 1.

ARBs Act as Non-Competitive Antagonists of Nb6e-Peptide Conjugates

We then asked how the inability of Nb6e-AT1R ligand conjugates to fully occupy the orthosteric ligand-binding site would affect their interaction with orthosteric ligands in functional assays. Pre-incubation with increasing concentrations of the ARB losartan caused in a rightward shift in the β-arrestin recruitment dose-response curves for AngII, TRV027, TRV055, and TRV056, as well as the G₃-Ahx-modified peptides, without altering the E_max (Fig. 7A, Supplemental Fig. 3A). Consistent with the expected pattern of competitive antagonism, Gaddum/Schild analysis of these data indicated a Schild slope close to 1 for all these ligands (Supplemental Table 5). However, losartan had little to no effect on the potency of the corresponding Nb6e-Ahx-peptide conjugates (Fig. 7B). Instead, increasing concentrations of losartan progressively lowered the efficacy of these ligands. A similar pattern of competitive and non-competitive antagonism by losartan was observed in Gq activation for AngII and Nb6e-Ahx-AngII, respectively (Supplemental Fig. 3, B and C). In contrast, pre-incubation with unconjugated Nb6e decreased the potency rather than the efficacy of Nb6e-Ahx-TRV055 in β-arrestin recruitment (Fig. 7C).

Fig. 7.

Inhibition of β-arrestin2 plasma membrane recruitment induced by orthosteric and dualsteric ligands. (A–C) Cells overexpressing 6e-AT1R were pre-treated with varying concentrations of the orthosteric antagonist losartan (A,B) or unconjugated Nb6e (C) for 15 minutes before stimulation with parental AT1R ligands (A) or Nb6e-Ahx-AT1R ligand conjugates (B,C) for 20 minutes. Data are normalized to the response induced by 1 μM AngII in the absence of antagonists and represent the mean ± S.D. of three independent experiments each performed with two technical replicates. Parameters derived from (A) Gaddum/Schild or (B) operational model of allosterism curve fits are provided in Supplemental Tables 5 and 6, respectively. (D) Summary of the varying effects that conjugation of Nb6e has on the pharmacological properties of AT1R ligands.

Losartan’s behavior with Nb6e conjugates is consistent with that of a non-competitive antagonist, and these data could be fit using an operational model of allosterism (Leach et al., 2007) (Supplemental Table 6). The nanobody-receptor interaction is required for losartan to display a non-competitive inhibition pattern, as losartan acts as a competitive antagonist of Nb6e-Ahx-AngII at wild-type AT1R (Supplemental Fig. 3D, Supplemental Table 5). The decreased efficacy of Nb conjugates at 6e-AT1R in the presence of losartan also does not appear to reflect a state of hemi-equilibrium, since increasing the agonist stimulation time from 20 minutes to 50 minutes did not affect these findings (Supplemental Fig. 3, E and F). Notably, for all four ligand series, the Nb6e conjugates were less sensitive to losartan than their parental ligands, as indicated by the difference in the apparent affinities of losartan (Supplemental Tables 5 and 6).

Discussion

This study has shown that conjugation of the same allosteric nanobody to structurally similar but functionally diverse AT1R ligands modulates their pharmacological properties in a ligand-specific manner (Fig. 7D). While all of the resulting dualsteric ligands had enhanced potencies compared with the G₃-linker-modified peptides used for sortase ligation, only two of the four (Nb6e-Ahx-TRV055 and Nb6e-Ahx-TRV056) were more potent than their parental AT1R ligands (Fig. 2). These same two ligands showed substantial bias toward Gq activation versus β-arrestin recruitment in comparison with the parental ligands (Fig. 3). A different pair of dualsteric ligands, Nb6e-Ahx-TRV027 and Nb6e-Ahx-TRV056, were less efficacious than their respective parental AT1R ligands (Fig. 2). All four dualsteric ligands were inhibited by losartan in a non-competitive manner (Fig. 7B), including Nb6e-Ahx-AngII, which otherwise performs similarly to AngII in functional assays. This pattern of inhibition occurs even though the functionally active peptide binds to the same orthosteric site as ARBs, and we directly demonstrated that Nb6e conjugates cannot fully compete with a radiolabeled ARB (Fig. 4). This underscores the fact that dualsteric ligands can exhibit pharmacological characteristics of both orthosteric and allosteric ligands.

We have also observed differential effects of Nb6e conjugation on the activity and bias of two ligands of the neurokinin-1 receptor (Braga Emidio and Cheloha, 2023), another peptidergic Family A GPCR with a comparably sized N-terminus (27 amino acids for neurokinin-1 receptor versus 23 for AT1R, as defined by generic GPCR numbering (Ballesteros and Weinstein, 1995; Isberg et al., 2015)). Nb-ligand conjugates targeting parathyroid hormone receptor-1, a class B1 GPCR with a large extracellular domain, display potent and unexpected signaling properties (Cheloha et al., 2020a, 2021). For all these receptors, orthosteric ligands with relatively low affinities tended to show the most profound changes in activity following nanobody conjugation, as seen in the AT1R system for TRV055 and TRV056. Parathyroid hormone receptor-1 differs from AT1R and neurokinin-1 receptor in that the N-terminal portion of its peptidic ligand inserts into the transmembrane orthosteric site, requiring the Nb-ligand linker to be placed at the peptide C-terminus. The successful development of Nb-ligand conjugates with linkages at either the ligand N- or C-terminus highlights the adaptability of the dualsteric approach described here.

Compared with these previous studies, the larger panel of ligands explored here shows that the combination of properties affected by Nb6e conjugation is not readily predictable a priori. For example, for the two Gq-biased peptides, we might have predicted that the shorter TRV055 would be more sensitive to modification than TRV056 (Fig. 1E). AngII analogs bind to the receptor with the C-terminus at the base of the binding pocket and the N-terminus oriented toward the cytoplasm (Wingler et al., 2019b; Wingler et al., 2020). Despite the N-terminal extension of TRV056, which could impart more flexibility to allow proper engagement of the peptide moiety, Nb6e-Ahx-TRV056 but not Nb6e-Ahx-TRV055 had reduced efficacy in functional assays. Until we have a fuller understanding of the mechanistic reasons for these differences, each conjugate must be tested empirically.

Altering the linker between the allosteric and orthosteric moieties of dualsteric ligands can often have profound effects on their pharmacology (Tahtaoui et al., 2004; Narlawar et al., 2010; Campbell et al., 2017;). Optimal linkers will favorably position both binding epitopes for receptor interaction without introducing a high entropic penalty from conformational flexibility (Mammen et al., 1998; Mohr et al., 2010). Steric constraints imposed by linkers could also alter the binding mode of each epitope (Guo et al., 2014) and thus fundamentally change their effects on receptor activation. Yet in our system, varying the linker length over a range of ∼18 Å did not substantially affect the potency or efficacy of either the full agonist Nb6e-AngII series or the partial agonist Nb6e-TRV027 series (Figs. 5 and 6). The pharmacology of these dualsteric ligands was determined principally by the identity of the orthosteric ligand rather than the nature of the linkage. When developing dualsteric ligands for certain systems similar to ours, screening a variety of orthosteric ligands might be more productive than exhaustively screening linkers.

Small molecule allosteric GPCR ligands typically have lower affinity than orthosteric ligands, which bind to a well-defined pocket that has been evolutionarily optimized for ligand binding. Even though unconjugated Nb6e acts as a neutral or negative allosteric ligand (Fig. 1D, Supplemental Tables 2 and 3), the high affinity of Nb6e enables it to increase the effective concentration of conjugated orthosteric moieties and thus ligand potency—a property associated with positive allosteric modulators. Nanobodies lend themselves to the development of allosteric ligands with such bespoke properties. Using either classic immunization-based approaches or purely in vitro methods, nanobodies can be discovered on the basis of GPCR binding, facilitating the discovery of even functionally silent, neutral allosteric ligands. Nanobodies can also be evolved to reach very high affinities (Pardon et al., 2014; McMahon et al., 2018; Zimmermann et al., 2018; Wellner et al., 2021) and targeted to defined regions or epitopes. Their epitopes can include less-structured regions that small molecules would be unable to bind, such as the N-termini of GPCRs. Since GPCR N-termini are often subject to alternative splicing (Oladosu et al., 2015; Marti-Solano et al., 2020; Schihada et al., 2022), isoform-specific dualsteric ligands could be an intriguing way to target receptors in particular cell types.

Our study leveraged an existing nanobody/epitope tag combination to develop dualsteric ligands for the AT1R. While such ligands cannot be used to target endogenous receptors, we propose that this type of engineered system has utility for exploratory purposes. For example, before investing substantial resources to develop a receptor-specific nanobody for use in dualsteric ligands, epitope tags could be placed at multiple sites to determine the optimal positioning, and nanobodies could be raised against that region of the GPCR. Dualsteric ligands could be targeted to specific cell populations in vivo by regulating expression of an epitope-tagged GPCR through cell-type-specific promoters or Cre-lox recombination technology, by analogy to the synthetic-ligand-specific Designer Receptors Exclusively Activated by Designer Drugs (Roth, 2016). Worth noting in this context, Nb6e-Ahx-TRV055 exhibits >1,000-fold selectivity for tagged versus untagged receptors (Fig. 2, I and J, Supplemental Table 1), suggesting that highly specific in vivo targeting is feasible.

In addition to highlighting the varied pharmacologies dualsteric AT1R ligands can achieve, our data reveal a novel mechanism to bias AT1R signaling toward Gq. Recent biophysical and computational studies from us and others have outlined a molecular mechanism that accounts for the clear structure-activity relationships of orthosteric AT1R biased ligands (Wingler et al., 2019a; Suomivuori et al., 2020; Wingler and Lefkowitz, 2020; Wingler et al., 2020; Cao et al., 2023; Zhang et al., 2023). The Phe side chain of the C-terminal residue (Fig. 1E) is essential for Gq activation because it interacts with AT1R residues at the base of the orthosteric pocket, triggering a specific set of conformational changes (Wingler et al., 2019a; Wingler et al., 2020). Consistent with this model, Nb6e conjugation to the β-arrestin biased ligand TRV027 is not sufficient to confer Gq activity (Fig. 2E). Nevertheless, Nb6e conjugation biases TRV055 and TRV056 signaling toward Gq (Fig. 3, C and D). The enhanced allosteric coupling of the parental ligands TRV055 and TRV056 to Gq (Strachan et al., 2014) results from altered interactions within the orthosteric pocket. Specifically, these ligands lose electrostatic interactions between AngII Arg2 and Asp residues near the extracellular face of AT1R, opening up the orthosteric pocket to facilitate the ability of the peptide’s C-terminal Phe to assume its active conformation (Suomivuori et al., 2020). While it is possible that signaling bias of these conjugates results from steric constraints forcing the peptide moieties into an altered binding mode, the bias of Nb6e-Ahx-TRV055 and Nb6e-Ahx-TRV056 could be effected through mechanisms such as altered orthosteric ligand-binding kinetics (Klein Herenbrink et al., 2016; Lane et al., 2017).

In conclusion, using N-terminally epitope-tagged AT1R and a tag-specific nanobody as proof-of-principle, we demonstrated that bitopic ligands built from an allosteric nanobody and an orthosteric peptide ligand broaden the pharmacology of AT1R ligands. Compared with the original orthosteric ligands, the resulting dualsteric ligands tuned varying combinations of potency, efficacy, and signaling bias, and they all caused orthosteric antagonists to display properties of a non-competitive inhibitor. Notably, this spectrum of effects was achieved with a nanobody that acts as a neutral or negative allosteric ligand. This highlights the potential of the growing collection of GPCR nanobodies to serve as useful handles in dualsteric ligands, regardless of their intrinsic effects on GPCR activity.

Abbreviations

Ab

antibody

Ahx

6-aminohexanoic acid

AngII

angiotensin II

ARB

angiotensin receptor blocker

AT1R

angiotensin II type I receptor

CI

confidence interval

E_max

maximal efficacy

Fmoc

9-fluorenylmethyloxycarbonyl

GPCR

G protein-coupled receptor

K_d

dissociation constant

K_i

inhibition constant

Nb

nanobody

NTA

nitrilotriacetic acid

RP-HPLC

reversed-phase high-performance liquid chromatography

TFA

trifluoroacetic acid

Authorship Contributions

Participated in research design: Braga Emidio, Small, Cheloha, Wingler.

Conducted experiments: Braga Emidio, Small, Keller, Wingler.

Contributed new reagents or analytic tools: Braga Emidio, Cheloha.

Performed data analysis: Braga Emidio, Small, Wingler.

Wrote or contributed to the writing of the manuscript: Braga Emidio, Cheloha, Wingler.