G protein-coupled receptors (GPCRs) represent the single most abundant and targeted class of druggable proteins in the human genome1,2. GPCRs sense an immense array of intra- and extracellular stimuli including neurotransmitters, metabolites, odorants, and drugs2, and allosterically modulate a variety of transducers including as many as 16 Gα proteins3 and 4 arrestins to control downstream cellular signaling4. It is now recognized that different GPCR agonists and antagonists can stabilize distinct receptor conformations with varying potencies and efficacies thereby differentially activating downstream signaling pathways. The process by which ligands differentially modulate cellular signaling is known as functional selectivity5, and ligands that display this property are referred to as biased. Recent advances in spectroscopic and crystallographic studies have facilitated the observation of these ligand-specific states supporting the existence of an array of GPCR conformational states6,7,8. The molecular mechanisms by which ligands stabilize these GPCR conformational changes are, however, incompletely understood. A key to understanding these transition mechanisms is to obtain structures of each state, not merely generic inactive and active states.

Nanobodies (Nbs)9 potentially provide a powerful platform to stabilize specific receptor states. As single-domain antibodies developed from camelids, nanobodies display unique properties such as small-size and high-affinity. Nanobodies have been widely used to stabilize the active states of many GPCRs, including the β2- adrenergic (β2AR), M2-muscarinic (M2R), µ-opioid (MOR), k-opioid (KOR), and angiotensin II type 1 (AT1R) receptors10,11,12,13,14,15. Apart from the recognition and stabilization of specific conformations, nanobodies also act as allosteric effectors and can affect both ligand binding affinity and functional activity8. For example, stabilization of an active and inactive conformation and dramatic alterations in ligand affinity at the β2AR by allosteric nanobodies were recently reported8. These state-specific nanobodies have also been widely applied as conformational biosensors to monitor GPCR dynamics both in vitro and in vivo16,17,18,19. The available structures of receptor–nanobody complexes reveal that the specific interfaces frequently overlap, at least in part, with G protein interaction surfaces20.

We previously reported two allosteric nanobodies against KOR—a positive allosteric nanobody (Nb39), and a negative allosteric nanobody (Nb6)14. The KOR has emerged as an important therapeutic target because G protein biased KOR agonists are reported to possess analgesic activity without the liabilities of addiction and respiratory depression associated with conventional opioid analgesics which target the μ receptor21,22,23,24. As well, KOR antagonists and KOR negative allosteric modulators have been proposed both as potential antidepressants and as treatments for addiction25,26,27. To date, no bona fide allosteric binding sites have been identified in KOR, nor are allosteric modulators available for KOR.

To systematically probe the activities of these nanobodies and determine how they reciprocally regulate ligand binding, we investigate the allosteric range of KOR in the absence and presence of Nb6 and Nb39. While Nb39 stabilizes KOR in a fully active state similar to MOR stabilized by Gi heterotrimer14,28, we show here that Nb6 stabilizes an inactive state distinct from that stabilized by a neutral antagonist. To better understand where Nb6 binds to the receptor and how it allosterically modulates the ligand binding, we solve the complex structure of the KOR bound to the inverse agonist JDTic in the presence of Nb6. Analysis of the KOR–Nb6 complex reveales a unique nanobody binding interface by which Nb6 negatively allosterically modulates KOR. Remarkably, we find that this site can be transferred to several other GPCRs to provide an engineered and functional allosteric site. Finally, given that Nb6 and Nb39 interact with distinct states of KOR, we report that they can be used as biosensors to detect ligand-stabilized states in real-time with subcellular resolution in living cells.


Molecular pharmacology of allosteric nanobodies

Previously we reported via a Bioluminescence Resonance Energy Transfer (BRET) approach that KOR agonists promote the association of Nb39 and the dissociation of Nb6 from KOR14. A simplified model depicting how agonists affect receptor-nanobody interaction is shown in Fig. 1a. The initial net BRET ratio between KOR and Nb6 in the absence of agonist is high (Fig. 1b), implying that Nb6 preferentially binds to the inactive state of KOR. After exposure to high concentrations of the KOR selective agonists Salvinorin A (Sal A)29 or Dynorphin A (1–17), conformational changes are stabilized that are no longer favorable for Nb6 binding. Consequently, Nb6 apparently dissociates from the receptor. As also shown in Fig. 1b the KOR selective antagonist JDTic30 modestly potentiates Nb6 binding as might be predicted from the fact that Nb6 already has stabilized the inactive state. In Fig. 1c, the agonist-induced active state of KOR is recognized by Nb39 as evidenced by enhanced nanobody binding as monitored by BRET. We note that SalA and Dynrophin A display different efficacies for Nb39 with SalA being considerably more efficacious than Dynorphin A. Two factors may account for this observed differential efficacy for Nb39 recruitment between Sal A and Dynorphin A: the first is that the receptor state or conformation is ligand-specific8 (Dynorphin A may induce an active conformation which is not favorable for Nb39 recognition); the second is that cellular accessibility to the two agonists are different. Pertinent to this second possibility we note that Sal A as a diterpene could quickly diffuse across the membrane and activate intracellular receptors (as supported by confocal imaging in this manuscript), while Dynorphin A as a 17 amino acid peptide, would inefficiently penetrate the cell membrane leading to lesser interactions with intracellular receptors.

Fig. 1: Molecular pharmacology of Nb6 and Nb39.
figure 1

a Schematic of Nb6 and Nb39 interaction with KOR in BRET-based assay. b Agonist-induced dissociation of Nb6 from the KOR detected by BRET assay. EC50: Sal A (10.2 ± 0.15 nM); Dynorphin A (155.6 ± 28.6 nM). Note that concentration at −12 represents no-drug added. Net BRET ratio was obtained by the signal(KOR+Nb6) minus signal(KOR). Each data point represents a mean ± s.d. in error bars (N = 5, five experiments each done in duplicate) c Agonist-induced association of Nb39 with the KOR detected by BRET assay. EC50: Sal A (6.04 ± 0.54 nM); Dynorphin A (1.38 ± 0.17 nM). Each data point represents a mean ± s.d. in error bars (N = 5, five experiments each done in duplicate). d Saturation binding of the agonist radioligand 3H-U69,593 at KOR in the presence of Nb6 or Nb39. Bmax value from the saturation curve was obtained by fitting to the one-site saturation model in Graphpad Prism. c.p.m., counts per minute. Bmax value: 1288 for KOR untreated; 455 for KOR+Nb6; 4880 for KOR+Nb39. (N = 3, three experiments each done in duplicate). eg Binding affinity of the antagonist JDTic (e), endogenous agonist Dynorphin A (f) and neutral antagonist LY2459989 (g) in the presence of Nb6 or Nb39. The statistical values are quantified in Supplementary Table 2 by fitting to the “one site-fit Ki” program in GraphPad Prism. Error bars represent SEM. (N = 3, three experiments each done in duplicate). Source data are provided as a Source Data file.

To further quantify the apparent allosteric effects of Nb6 and Nb39, multiple assays were conducted to examine how these nanobodies differentially regulate KOR ligand binding and functional activity. First, saturation binding was conducted using a KOR selective agonist radioligand 3H-U69,593 in the presence of Nb6 or Nb39. Based on the extended ternary complex model of GPCR activation31, the coupling of G protein or its mimetic (e.g., a nanobody) increases the number of high-affinity agonist binding sites8,13,14. Here we demonstrated that Nb6 reduces 3H-U69,593 binding by stabilizing an inactive state of KOR, which is unfavorable for agonist binding, while Nb39 increases radioligand binding by stabilizing an active-like state (Fig. 1d). Secondly, a series of KOR ligands with distinct scaffolds were tested in competition binding assays in the presence of each nanobody to investigate the allosteric range of the two state-specific nanobodies. For the competition binding assays, the radioligand 3H-U69,593 is no longer a suitable radiotracer for this assay because, as mentioned earlier, its binding kinetics are also affected by the nanobodies. Another available antagonist radioligand 3H-JDTic was also tested, and no significant changes were detected for its binding affinity (Ki) or equilibrium constant (Kd) with KOR in the presence of Nb6 or Nb39 (Fig. 1e, Supplementary Fig. 1a and Fig. 2); 3H-JDTiC was thus suitable the studies reported here. For the endogenous ligand Dynorphin A (1–17), a 200-fold higher affinity for KOR in the presence of Nb39 compared to the affinity in the presence of Nb6 was observed (Fig. 1f).

Fig. 2: Overall alignment of KOR–Nb6 and KOR–JDTic and KOR–Nb39 structures.
figure 2

a Overall architecture of KOR–JDTic–Nb6 complex. The receptor is shown in green, Nb6 in purple, and JDTic in orange. b, c Extracellular (b) and intracellular (c) views of structural alignment of KOR–Nb6 complex (green) and KOR–JDTic inactive state (gray; PDB ID code 4DJH). d, e Extracellular (d) and intracellular (e) views of a structural alignment KOR–Nb6 complex (green) and KOR–Nb39 active state (blue; PDB ID code 6B73). Major structural differences are shown as red arrows indicate.

Unexpectedly, the binding affinity of a KOR neutral (or silent) antagonist LY245998932 was increased ~6-fold in the presence of Nb6 (Fig. 1g). This phenomenon is reminiscent of a prior observation at the β2AR where Nb60 and Nb80 promoted a full allosteric range for agonists from very-low-affinity to high-affinity states, respectively. Apo (unliganded) receptors exist in multiple inactive states that exchange quickly and thus display conformational heterogeneity8. Similarly, for a receptor bound only to an agonist, multiple states such as inactive, intermediate, and active states have been observed, and more complete stabilization of the active state (high-affinity) requires the engagement of transducers (e.g., G proteins or its mimimetic) and agonists6,7,33. The increased affinity of LY2459989 supports this observation that the addition of Nb6 to KOR helps stabilize the receptor into a state which is favorable for LY2459989’s binding. We also determined the binding of agonist ligands of diverse scaffolds, including the peptide Dynorphin A (1–13), the diterpene Sal A, the epoxy-morphinan MP1104, and the arylacetamide GR89696. As shown in Supplementary Fig. 1b, all displayed the expected effects induced by the positive (Nb39) and negative (Nb6) allosteric nanobodies: their affinities were attenuated by the negative allosteric nanobody Nb6 and enhanced by the positive allosteric nanobody Nb39. One exception was that the agonist MP1104’s binding affinity was only slightly affected by Nb6 compared to other agonists, which emphasizes that the allosteric effect is also probe-dependent34. MP1104 has displayed very slow off-rate14 and promotes a specific receptor conformation, which is no longer favorable for Nb6 binding.

Finally, the functional activity of several KOR agonists or partial agonists was evaluated in the presence of Nb6; we previously reported that Nb39 sterically inhibits G protein binding and enhances β-arrestin recruitment14. Here we find that Nb6, consistent with its ability to stabilize an inactive state, decreases the potency and/or efficacy of selected KOR agonists (Supplementary Fig. 3). Together, these data reveal that Nb6 and Nb39 are potent allosteric nanobody modulators that can be used to stabilize inactive and active KOR states. Furthermore, these states can be distinguished from the apo state by a variety of criteria.

Structure of the KOR–JDTic–Nb6 complex

As Nb6 promotes a unique receptor state distinct from either the neutral-antagonist-bound or Nb39-stabilized states, we asked if the Nb6-stabilized structure differs from previously reported inactive- and active-like KOR structures14,35. To crystallize the KOR–Nb6 complex, the same N-terminal BRIL fusion constructs like the one previously engineered to obtain KOR–Nb39 active-state structure14 was used to avoid any modification at the intracellular regions of the receptor. Multiple KOR antagonists were then screened in thermostability assays for their ability to stabilize the complex. Of those tested, the antagonist JDTic36 afforded maximal stabilization of the KOR–Nb6 complex (Supplementary Fig. 4a) as monitored by size-exclusion chromatography. We found JDTic to be a potent inverse agonist at KOR (Supplementary Fig. 4b) via a generic assay which measures the Gi-mediated inhibition of cAMP production. With more specific BRET-based assays, JDTic appears to differentially decrease the coupling between constitutively active KOR and specific Gi/o family members, with greatest efficacy for Gz, GαoA, and GαoB proteins (Supplementary Fig. 4c). As JDTic promotes a stable KOR–Nb6 complex, the X-ray structure of KOR–JDTic–Nb6 was determined to 3.3 Å resolution in the space group P21221 with two monomers per asymmetric unit (Fig. 2a, Supplementary Fig. 4d and Supplementary Table 1). This observed anti-parallel crystal packing should be attributed to the unique crystallization conditions, and is unlikely to represent possible dimerization under physiological conditions. We observed strong densities for KOR, Nb6, and JDTic (Supplementary Fig. 4e). No electron density was observed for the presumably disordered N-terminal BRIL fusion, as was the case for the active state KOR structure14. Furthermore, because the two monomers (chain A and B) are nearly identical, we focused on the chain A complex for subsequent analysis.

Together with the previously reported KOR–JDTic (nanobody free)35 and KOR–Nb3914 structures, we next analyzed the overall helical movements by comparing them with the KOR–Nb6 structure reported here. The overall structure of KOR–Nb6 is reminiscent of the inactive-state KOR–JDTic structure (Fig. 2b, c) although substantial differences were observed at both extracellular [transmembrane helical (TM) 1 and 5, extracellular loop (ECL) 2 and 3] and intracellular regions [TM5, TM6 and intracellular loop (ICL) 2]. Significantly, ECL2 moves ~6 Å to form an apparent lid on top of JDTic, a feature recently observed in other GPCR structures, including the LSD-bound 5-HT2B serotonin receptor37 and the risperidone-bound D2 dopamine receptor38. The formation of this lid may explain why the binding kinetics (e.g., kon and koff) in KOR preincubated with Nb6 were slightly changed (t1/2 increases ~2-fold) compared to that in the untreated KOR (Supplementary Fig. 1a). It is also noteworthy that interactions between Nb6 and TM5/TM6 facilitate a mild inward movement of TM6 (~2 Å) by measuring the distance between Cα atoms of D2666.27 (superscript denotes Ballesteros–Weinstein numbering). The overall structural similarity between the JDTic-only and Nb6-bound forms is likely because the inverse agonist activity of JDTic stabilizes the receptor in a low-affinity state for agonists, as the binding affinity of JDTic is nearly identical with or without Nb6 (Fig. 1d).

The structural alignment between KOR–Nb6 and active-state KOR–Nb39 revealed canonical active state-like conformational changes and side-chain displacements in the binding pocket and various conserved motifs (Supplementary Fig. 5). Specifically, an inward movement of extracellular TM1, TM5, TM6, and TM7 (Fig. 2d) leads to the contraction of the binding pocket, which is a joint event upon GPCR activation, as also observed in several other Class A GPCR crystal structures10,11,13,14. Compared with the active-state KOR, the conformation of ICL3 undergoes significant displacement due to the large outward movement of TM6 (Fig. 2e). Similarly, displacement of intracellular TM5 and TM7 in the active-state KOR is likely a direct effect of TM614, and the role of TM6 as a driving force for the expansion of the intracellular part indicates that its conformational differences may play an essential role in determining receptors’ coupling specificity for G proteins39.

Unique interactions revealed between Nb6 and KOR

As shown in Fig. 3a, the Complementarity-determining region 3 (CDR3) loop of Nb6 inserts into a crevice between TM5 and TM6, where it forms major interactions with KOR. This interface is distinct from previously reported nanobody-GPCR interfaces (e.g., β2AR-Nb60, β2AR-Nb80, M2R-Nb9-8, MOR-Nb39) and the active-state-like KOR–Nb39 interfaces (Supplementary Fig. 6a), all of which bind via the intracellular core region which is also critical for G protein11 and/or β-arrestin40 interactions. From a nanobody perspective, even though the CDR3 adopts a flexible loop conformation, the electron density in this region was sufficient to identify the orientation of side-chains within the residues (Supplementary Fig. 6b). For instance, W106 in Nb6 extends into the crevice and forms hydrophobic interactions with L2535.65 and the backbone of R2746.35, likely anchoring TM5 and TM6, and preventing the latter from moving outward. R102 in Nb6 forms ionic interactions with two charged residues (D2666.27 and R2706.31) in the receptor, providing an H-bond network linking TM5 and TM6. All of these interactions appear to contribute to the high affinity of the KOR–Nb6 interaction because mutations of these residues completely abolish the binding of Nb6 to the KOR in BRET assays (Fig. 3b, Supplementary Fig. 6c). Interestingly, Y34 and I105 in Nb6 do not directly interact with the receptor, but mutations of these two residues also attenuate the interaction likely because Y34 and I105 interact with R102 and W106, respectively, assisting the latter two residues to be in the right orientation (Fig. 3b, c). Although mutation of several other residues (L2535.65 and R2746.35) appears not to affect Nb6 binding to the receptor apparently, they do change the stability of KOR/Nb6 complex as the decreased net BRET ratio and increased potency of agonist Sal A indicate that a weaker interaction between Nb6 and KOR has been formed (Supplementary Fig. 6c).

Fig. 3: Unique interface between KOR and Nb6.
figure 3

a Details of the interface between KOR (green) and Nb6 (purple) are shown. b Mutagenesis studies using BRET identify key residues essential for KOR–Nb6 interactions. Error bars represent SEM. (N = 3) c left, ionic-lock in β2AR stabilizing the S2 inactive state; middle, R3.50 forms two hydrogen bonds with T2.39 and T6.34, respectively. These interactions mimic the ionic lock observed in β2AR; right, R102 in Nb6 mediates a new H-bond network between TM5 and TM6. This interaction is not observed in either KOR–JDTic inactive or KOR–Nb39 active states. Source data are provided as a Source Data file.

Although the overall differences between KOR–Nb6 and KOR–JDTic are subtle, there are some changes specific to the Nb6-stabilized state. Apart from the large movement of the ECL2, a twist of the intracellular region of TM5 and TM6 has been observed in the KOR–Nb6 structure (Supplementary Fig. 7a). This twist causes large conformational displacements of side chains of residues in the interface engaged by the CDR3 loop in Nb6 (Supplementary Fig. 7b). The relatively differential position of the TM5 and TM6 also supports that the KOR–Nb6 structure may represent a distinct state from previous KOR–JDTic because the dynamics of TM5 and TM6 has been seen as a major event during receptor transition from inactive to active states11,28,40. Because of the stabilization afforded by Nb6, we were able for the first time to visualize an intact ICL3 (i.e., ICL3 is not visualized in previous KOR–JDTic or KOR–Nb39 structures). A comparison of ICL3 between KOR–Nb6 and MOR-Gi structures reveals a significant translocation, probably necessitated accommodating the Gi protein coupling (Supplementary Fig. 7c). Since Nb6 is a negative allosteric modulator and its binding pocket represents a potentially allosteric site, we continued to investigate how the residues located in the binding pocket contribute to the KOR signaling profile. While alanine mutagenesis shows that most of the residues have minor effects on KOR’s G protein activation and β-arrestin translocation, L2535.65A and R2716.32A appear to significantly reduce receptor activity (Supplementary Fig. 7d). In particular, R2716.32A affects the G protein signaling but not arrestin recruitment indicating this residue may play a unique role in KOR-G protein interaction. This is supported by the observation in the closely related MOR-Gi structure that R2776.32 in MOR forms an H-bond interaction with the backbone of L353G.H5.25 in Gαi1 proteins28. The R/K6.32 is conserved between different GPCR families: in β2AR, K6.32 forms important interactions with Gαs39 while in class F GPCRs, R6.32 acts as a conserved molecular switch to regulate receptor activation and pathway selection41. The L2535.65A mutation affects both G protein and arrestin signaling simultaneously, indicating that it may play a global role in stabilizing the TM5-TM6 interaction in the active state. Overall, the KOR–Nb6 interface represents an allosteric site distinct from either the G protein or β-arrestin binding site, which may have the potential in the design of KOR allosteric modulators.

Regarding the negative allosteric mechanism of Nb6, we note that in the previously described inactive-state β2AR-Nb60 structure8, residue T103CDR3 in Nb60 acts as a bridge forming an ionic lock between D3.49 in TM3 and E6.30 in TM6 of β2AR (Fig. 3c, left). This ionic lock has been found conserved in several aminergic GPCRs in their inactive states; when broken this ionic lock appears to be one of the hallmark events of receptor activation8,42. However, there is no such ionic bridge observed in opioid receptors as there is a Leu6.30 in all opioid receptors, indicating a different mechanism may be involved in the Nb6-mediated KOR inactivation. Here, R3.50 in KOR forms two H-bonds with T2.39 in TM2 and T6.34 in TM6, respectively (Fig. 3c, middle), thereby stabilizing the inactive conformation. Further, R102 in Nb6 forms a hydrogen bond with D2666.27, resulting in a new H-bond network formed by D2666.27, R2706.31, and L2535.65 between TM5 and TM6 (Fig. 3c, right). These interactions jointly stabilize an inactive state of KOR by potentiating the interaction between TM5 and TM6 and, importantly, were not observed in the KOR–JDTic or KOR–Nb39 structures. Taken together, these observations suggest that Nb6 stabilizes a different inactive state through combined effects of steric hindrance by W106 and ionic locking by R102 as mutations of these residues abolishes Nb6 binding. Mutations of the critical residues on the interface (R102 or W106 in Nb6) also abolish its inhibitory action for Sal A-mediated cAMP inhibition (Supplementary Fig. 8a).

Nb6/39 as bidirectional biosensors for opioid receptors

As shown earlier, Nb6 and Nb39 interact with distinct, non-overlapping receptor conformations. These discrete interaction profiles are potentially ideal for the development of biosensors to probe ligand-stabilized receptor conformations in real-time. To test the applicability of using Nb6 and Nb39 as KOR conformational biosensors, we first used the BRET assay to determine if the Nb6 or Nb39 interactions with KOR were reversible. For these studies, KOR-Rluc (Renilla Luciferase) and Nb6-mVenus or Nb39-mVenus were co-expressed in HEK 293T cells. We first tested the ability of Nb6, the inactive-state associated nanobody, to report changes in receptor activity in vitro. The agonist Sal A caused a reasonably rapid (t1/2 = 50 ± 10 s) decrease in Nb6-KOR BRET (Fig. 4a), which plateaued after ~10 min. The subsequent addition of the antagonist JDTic rapidly (t1/2 = 75 ± 12 s) increased the BRET ratio equivalent to the level without agonist treatment. JDTic has a much higher binding affinity than Sal A at wild type KOR (Kd, 0.065 nM for JDTic vs 10.9 nM for Sal A) (Fig. 1d, Supplementary Fig. 1b). JDTic also has a slower off-rate in the presence of Nb6. Conversely, Nb6 coupling destabilizes Sal A binding and thus accelerates the dissociation of the agonist Sal A, as supported by the lower Kd measured compared to the KOR/buffer group (Supplementary Fig. 8b)

Fig. 4: Nb6 and Nb39 as biosensors probe KOR conformation in real-time.
figure 4

a, b Association and dissociation kinetics of KOR–Nb6 (a) or KOR–Nb39 (b). The experiments were conducted using BRET-based assays and data were analyzed by GraphPad Prism. 10 µM agonist Sal A or antagonist JDTic was added at indicated time points by black arrows. (N = 5, five experiments each done in duplicate) c Confocal live-cell imaging of HEK 293T cells expressing KOR-mScarlet and Nb6-mVenus. 10 µM Sal A or JDTic was added to the cell plate at indicated time points. (N = 3, three experiments each done in duplicate) d Confocal live-cell imaging of HEK 293T cells expressing KOR-mScarlet and Nb39-mVenus. 10 µM Sal A or JDTic was added at indicated time points. Error bars represent SEM. (N = 3, three experiments each done in duplicate). Source data are provided as a Source Data file.

We next evaluated the ability of Nb39, the active-state associated nanobody, to report real-time changes in ligand-stabilized states in live cells in vitro. Here the agonist Sal A immediately enhanced the BRET signal between Nb39 and KOR (Fig. 4b). JDTic rapidly (t1/2 = 107 ± 15 s) reversed the KOR–Nb39 association, presumably by stabilizing the inactive state (Fig. 4b). The relatively slower t1/2 for JDTic in KOR/Nb39 (~107 s) compared to KOR/Nb6 (~75 s) is likely due to the cooperativity from Nb39 because Nb39 has been shown to significantly increase the binding affinity of Sal A (Supplementary Fig. 8b). This reversible interaction between receptor and nanobody could provide a potential conformation-specific biosensor approach for investigating these transitions for a variety of applications.

We next investigated the association of Nb6 and Nb39 with KOR using confocal microscopy to determine if these ligand-stabilized transitions can be visualized in live cells in real-time with subcellular resolution. Here the distribution in single cells was monitored using KOR-mScarlet and Nb6- or Nb39-mVenus. For the first experiment, KOR-mScarlet and Nb6-mVenus are initially co-localized, as predicted from our previous observations that Nb6 binds ligand-free KOR (Fig. 4c). After treatment with Sal A, Nb6 dissociated from the plasma membrane (where it had been co-localized with KOR) and displayed increased cytosolic localization (Fig. 4c, Supplementary Fig. 9a, b). Subsequent treatment with the KOR selective antagonist JDTic reversed this phenomenon.

Next, when KOR-mScarlet and Nb39-mVenus were visualized, the Nb39-mVenus fluorescence was mainly found in the cytoplasm before drug treatment. This indicates that Nb39 recruitment to plasma membrane receptors does not occur in the absence of agonist (Fig. 4d, left panel) and is consistent with our BRET studies which show minimal basal BRET. Upon Sal A stimulation, Nb39 fluorescence rapidly accumulated at the plasma membrane to be co-localized with KOR-mScarlet (Fig. 4d, middle panel). In addition to the Nb39 localization to the plasma membrane, we also observed high co-localization of Nb39 and KOR in discrete subcellular compartments in the cytoplasm (Fig. 4d, Supplementary Fig. 9c, d), reminiscent of the Golgi apparatus19. Indeed, in cells expressing the Golgi marker GalT-RFP (Supplementary Fig. 9e, f), following Sal A treatment Nb39-mVenus rapidly localized to this GalT-RFP cellular compartment. The apparent translocation occurs within a few seconds following Sal A addition, consistent with KOR resident under basal conditions in the Golgi apparatus (Supplementary Fig. 9f). Subsequent exposure to the antagonist JDTic partially reversed this, apparently inducing dissociation of Nb39 from KOR by stabilizing the inactive conformation (Fig. 4d, right panel; Supplementary Fig. 9e). These results indicate that these conformation-specific nanobodies can be used to report both cell-surface and intracellular ligand-stabilized receptor conformations in a time-dependent and reversible fashion. A potential advantage of using Nb6 to probe the inactive state is that, as Nb6 pre-binds to the ligand-free receptors, it allows us to establish a baseline signal followed by a ligand response signal necessary for accurately measuring the amplitude of change. Finally, we note that due to the highly conserved sequence and structure of opioid receptors (Supplementary Fig. 10a), Nb6 also binds to μ, δ and nociceptin opioid receptors with high affinity (Supplementary Fig. 10b). Nb6 thus can also report the activation and inactivation of the nociceptin receptor (Supplementary Fig. 10c), while Nb39 has been shown unable to bind to this receptor14. Thus, together with Nb39, this nanobody pair provides a toolset for studying the dynamics of conformational transitions in situ for other members of the opioid receptor family.

Nb6 as a potentially useful GPCR conformational sensor

Inspired by Nb6’s capability being as a biosensor for opioid receptors and its unique interactions between KOR and Nb6, we predicted that Nb6 might bind to other GPCRs to report receptor conformational transitions by making chimeric GPCRs. Our hypothesis (Fig. 5a) was that swapping out a portion of ICL3 from an arbitrary GPCR with ICL3 from KOR would provide a site for Nb6 binding. Considering that the interaction between Nb6 and KOR requires an intact ICL3, which has been implicated in playing an essential role in G protein/β-arrestin recognition11,40, we screened for optimal sites for KOR ICL3 insertion which would have a minimal effect on the activity of a generic Class A GPCR. We subsequently tested seven Class A GPCRs which interact with different G protein subfamilies: Gαs (DRD1, 5HT4R and HRH2), Gαi (NTS1R and 5HT5AR), and Gαq (5HT2AR and ETA). The full sequences for every chimeric receptor, after optimization, are provided in the Supplementary document. In Fig. 5b, we present data obtained with the Gαq-coupled human 5-HT2A serotonin (5-HT2AR) and ETA endothelin (ETA) receptors, both of which are important drug targets found in a subfamily of GPCRs quite distinct from opioid receptors. Both wild-type 5-HT2AR and ETA, did not display measurable interaction with Nb6 in the BRET assay (Fig. 5b). When their ICL3 regions were replaced by the KOR ICL3 (see details in Supplementary document), Nb6 was now able to bind to 5-HT2AR or ETA as monitored by BRET. Remarkably, their endogenous agonists (5-HT or endothelin-1) caused the dissociation of Nb6 from the chimeric 5-HT2AR or ETA receptors, respectively (Fig. 5b). Importantly, the chimeric receptors remain functional as quantified by G protein activation (Fig. 5c). Other chimeric receptors (D1-dopamine, 5-HT4-serotonin, H2-histamine, NTSR1-neurotensin and 5-HT5A serotonin) were investigated in a similar manner, and they all could respond robustly to respective endogenous ligand without significantly affecting G protein coupling (Supplementary Fig. 11 and 12).

Fig. 5: The chimeric approach allows Nb6 to bind to 5-HT2AR and ETA.
figure 5

a Schematic of GPCR-KOR chimera generation. b BRET assays showing that 5-HT2AR-KOR or ETA-KOR chimera can bind to Nb6. Agonist 5-HT or Endothelin-1 causes dissociation of Nb6 from the chimeric receptor 5-HT2AR or ETA, respectively. EC50 values were summarized in Supplementary Table 4. (N = 3, three experiments each done in duplicate). c BRET assays showing that chimeric receptor 5-HT2AR-KOR or ETA-KOR can activate canonical G protein coupling similarly compared to the wild type receptors. EC50 values were summarized in Supplementary Table 5. (N = 3, three experiments each done in duplicate). d A comparison between Nb6 and miniGq proteins in reporting activation and inactivation in 5-HT2AR or ETA in the presence of agonists and antagonists. Experiments were conducted in the BRET assay. The first and second arrow indicates the time when the agonists and antagonists were added, respectively. For different receptors tested, the agonist and antagonist were 5-HT/Risperidone for 5-HT2AR, and Endothelin-3/CI 1020 for ETA. The reason that Endothelin-3 was used for ETA here instead of Endothelin-1 is that Endothelin-1’s affinity is too high to be replaced off. e BRET assays show that Nb6 can also bind to ‘orphan’ GPCRs in their basal state. ***p < 0.001 as compared to wild type group, unpaired t test. Error bars represent SEM. (N = 4, four experiments each done in duplicate). Source data are provided as a Source Data file.

Given that other biosensors such as the miniG proteins have been frequently applied to monitor receptor activation43, here we also wondered if Nb6 could report receptor activation in as robust a manner as miniG proteins. Nb6 and miniG were tested in parallel to monitor receptor activation and inactivation in the presence of agonists and antagonists, respectively. As shown in Fig. 5d, Nb6 and miniGq perform comparably at detecting the active and inactive states of 5-HT2AR or ETA receptors. Interestingly, the miniGi occasionally failed to monitor receptor activation (e.g., 5HT5AR) by the endogenous ligand 5-HT, and the miniGs proteins were unable to report receptor inactivation by antagonists (Supplementary Fig. 13). Encouraged by this success, we created additional chimeric receptors with several understudied “orphan” GPCRs (e.g., GPR111, GPR149, and GPR151). Nb6 displayed robust interaction with these chimeric receptors in their basal state, while the wild type receptors do not associate with Nb6 (Fig. 5e). These examples and observations with seven different GPCRs imply that this chimeric approach could be used to faithfully interrogate ligand-stabilized states for the majority of Family A GPCRs for which such nanobodies are unavailable.


Here we show how two KOR state-specific nanobodies can be used as real-time reporters for monitoring both ligand-dependent and independent conformational states in vitro and in situ. To illuminate the mechanism by which one of these–Nb6–stabilizes an uniqueinactive state, we obtained the crystal structure of KOR bound to this inactive-state stabilizing nanobody. Together with the previously determined KOR structures with an antagonist JDTic and the Nb39-stabilized active state with MP1104, a comprehensive structural analysis of conformational changes related to KOR activation and inactivation was revealed. Finally, we provide a potential platform that affords real-time in vitro and in situ monitoring of ligand-stabilized GPCR conformations using Nb6 and chimeric GPCRs.

With regard to the conformational transitions of KOR, our data imply that the so-called unliganded or “apo” state as it exists in transfected cells likely consists of a mixture of active and inactive states—each of which can be differentially stabilized by conformation-selective nanobodies. Intriguingly, we find that Nb6 stabilizes a very low-agonist-affinity state reminiscent of that stabilized by the β2AR nanobody Nb608. Significantly, although they both stabilize a very low-affinity inactive state, they do so by interacting with entirely different receptor domains. Thus, for instance, an ionic lock formed in both the inactive β2AR and adenosine A2A receptors was reported to determine the conversion between S1 and S2 inactive states33,42. In the β2AR stabilized by Nb60, an S2 inactive state has been observed with an intact ionic lock mediated by Nb608. For opioid receptors, there is no such ionic lock, and as a consequence, Nb6 utilizes an entirely different mode for inactive state stabilization. This unique Nb6-KOR binding surface conceivably could provide a platform for the discovery of allosteric modulators by, for instance, automated docking approaches44. On the other hand, the active-state stabilizing nanobodies Nb39 and Nb80 have overlapping binding sites with G proteins and interact with the core of opioid and β2 adrenergic receptors. These findings imply that active-state stabilizing nanobodies might share a conserved mode of action.

The overall structural differences in KOR are subtle in the absence or presence of Nb6. This is likely due to the potent inverse agonist activity of JDTic, which can stabilize the receptor in a very low-affinity state, as supported by its nearly identical affinity in the presence and absence of Nb6. However, for the neutral antagonist LY2459989, Nb6 stabilizes a state which is more favored by LY2459989 compared with the ligand only state. This finding provides further support for the notion that even for the inactive KOR conformation, at least two states likely exist. How these two low-affinity agonist states might interconvert is currently unclear. Our observation that certain nanobody-stabilized states are favored by specific ligands (e.g., probe dependence) provides a potential opportunity for the identification of new-scaffold agonists and antagonists, as was recently proposed for other receptor–nanobody complexes45.

Our finding that merely swapping the cognate regions of KOR with several other GPCRs provides an approach for reporting ligand-induced conformations deserves further comment. Producing nanobodies typically requires purified GPCRs and suitable hosts (llamas or sharks) and then several months of additional work to identify and validate their activity9. Typically a mixture of nanobodies is recovered, which must be fully characterized as to specifically for the GPCR target and state. Alternatively, a more recent approach uses a yeast-based display of nanobody libraries, although here too, the effort involved is considerable—especially for identifying those which are state-specific15,46. In this paper, we show that for at least several unrelated GPCRs, merely swapping a few residues is sufficient to afford robust interactions in a ligand-modulated fashion. Further, given that this simple approach affords the ability to monitor conformational transitions in real-time in vitro and in situ provides the possibility that this could be a relatively general approach. As we have only tested this on Family A GPCRs we do not know if it can ultimately be ported to other GPCR families, although that is currently being investigated.

The technology—although useful—is not without potential confounds. We initially were concerned that this type of swap would alter the signaling ability of the chimeric receptors. In at least the seven cases investigated here, we have not been able to measure any significant degree of attenuation or potentiation of G protein interactions. Certainly, however, future users of the approach will need to verify that signaling, ligand recognition and so on are unaltered before proceeding further with the technology. We also note that the kinetics of ligand-induced alterations in nanobody interactions are rapid and that given the limitations of our current BRET-based readers, we were not able to calculate accurate kinetic constants for some of these interactions.

Finally, we note that KOR has emerged as a primary drug target for medication development for drugs devoid of the side-effects of conventional opioids; such drugs include nalfurafine and buprenorphine. How these drugs activate opioid receptors and produce differential cellular effects are still understudied due to a lack of specific tools. Recently Stoeber et al. used a nanobody-based biosensor with a different nanobody (Nb33) to investigate actions of different opioid drugs in neurons with subcellular resolution19. Ligand-specific activation patterns were revealed at a high level of spatiotemporal resolution, specifically between endogenous peptides and non-peptide ligands. The availability of both inactive and active nanobodies for KOR (Nb6 and Nb39, respectively) now enables us to look at a full dynamic range of opioid receptors in a variety of cellular environments. Additionally, our chimeric receptor approach potentially provides a robust and straightforward platform for investigating these phenomena with a large number of unrelated GPCRs.


Crystallization of KOR–JDTic–Nb6 complex

Thenanobodies used in this work were generated in Ilama and the detailed method has been illustrated in Che et al.9,14. Plasmids Nb6 and Nb39 were transformed to E. coli WK6 cells (provided by the Steyaert lab) and were then expressed in the periplasm and purified following steps 70–739. Nanobodies were concentrated and desalted to the buffer: 10 mM HEPES, 100 mM NaCl, and 10% Glycerol and stored at −80 °C for future use.

For crystallography trials, we used a fusion protein of human KOR with an amino-terminal apocytochrome b562 RIL (BRIL)14. The BRIL-KOR fusion protein was expressed and purified from Sf9 insect cells (Expression Systems, #94-001S). The purified receptor was incubated with 10 µM JDTic (final concentration) all the time during purification. The N-terminal 10× His-tag was removed by the addition of His-tagged TEV protease (Homemade) and incubation overnight at 4 °C. Protease, cleaved His-tag and uncleaved protein were removed by passing the suspension through equilibrated TALON IMAC resin (Clontech) and collected the flow-through. Excessive Nb39 (KOR/Nb39 molar ratio: 1:2) was then added to the protein sample and incubated for 3 h. KOR–JDTic–Nb6 complexes were then concentrated to ~30 mg/ml with a 100 kDa molecular weight cut-off Vivaspin 500 centrifuge concentrator (Sartorius Stedim).

KOR–JDTic–Nb6 complexes were reconstituted into the lipidic cubic phase (LCP) by mixing protein solution and a monoolein/cholesterol (10:1 w/w) mixture in a ratio of 2:3 v/v (protein solution/lipid) using the twin-syringe method47. Crystallization was set up in 96-well glass sandwich plates (Marienfeld GmbH) using 50 nl LCP drops dispensed from a 10 µl gas-tight syringe (Hamilton) using a handheld dispenser (Art Robbins Instruments) and overlaid with 1 µl of precipitant solution. Upon optimization, KOR–JDTic–Nb6 crystals were obtained in 100 mM Tris pH7.0, 360–400 mM Ammonium citrate dibasic, 28–32% PEG400. Crystals grew to a maximum size of 50 µm × 40 µm × 30 µm within three days and were harvested directly from the LCP matrix using MiTeGen micromounts before flash-freezing and storage in liquid nitrogen.

Data collection and structure solution and refinement

X-ray data were collected at the 23ID-B and 23ID-D beamline (GM/CA CAT) at the Advanced Photon Source, Argonne, IL using a 10 µm minibeam at a wavelength of 1.0330 Å and a Dectris Eiger-16m detector and Dectris Pilatus3-6m detector, respectively. Diffraction data were collected by exposing the crystals for 0.5 s to an unattenuated beam using 0.5° oscillation per frame. The resulting diffraction data from 31 crystals were indexed, integrated, scaled, and merged using HKL200048. Initial phases were obtained by molecular replacement in PHASER49 using 3 independent models of a truncated 7TM portion of KOR (PDB ID: 4DJH), a nanobody Nb6 from the KOR–MP1104–Nb39 complex (PDB ID: 6B73), and the thermostabilized apocytochrome b562RIL protein (PDB ID: 1M6T)50. Two copies of the 7TM portion of each the receptor and the nanobody Nb6 but no BRIL were found in the asymmetric unit. Refinement was performed with PHENIX51 and REFMAC552 followed by manual examination and rebuilding of the refined coordinates in the program COOT53 using 2mFo - DFc and mFo - DFc maps. The data collection and refinement statistics are shown in Supplementary Table 1. The structural alignment and other analyses were performed in Pymol.

Bioluminescence Resonance Energy Transfer (BRET) assay

To measure KOR-nanobody recruitment, HEK 293T cells (ATCC, #CRL-11268) were co-transfected in a 1:3 ratio with human KOR containing C-terminal Renilla luciferase (RLuc8) and nanobody containing a C-terminal mVenus. After at least 16 h, transfected cells were plated in poly-lysine coated 96-well white clear bottom cell culture plates in plating media (DMEM + 1% dialyzed FBS) at a density of 25–50,000 cells in 200 µl per well and incubated overnight. The next day, media was decanted and cells were washed twice with 60 µL of drug buffer (20 mM HEPES, 1X HBSS, pH 7.4), then 60 µL of the RLuc substrate, coelenterazine h (Promega, 5 µM final concentration in drug buffer) was added per well, incubated an additional 5 min to allow for substrate diffusion. Afterward, 30 µL of the drug (3×) in drug buffer (20 mM HEPES, 1X HBSS, 0.1% BSA, pH 7.4) was added per well and incubated for another 5 min. Plates were immediately read for both luminescence at 485 nm and fluorescent mVenus emission at 530 nm for 1 s per well using a Mithras LB940 multimode microplate reader. The ratio of mVenus/RLuc was calculated per well and the net BRET ratio was calculated by subtracting the mVenus/RLuc per well from the mVenus/RLuc ratio in wells without Nb-mVenus present. The net BRET ratio was plotted as a function of drug concentration using the log(agonist) vs. response model in Graphpad Prism 5 (Graphpad Software Inc., San Diego, CA). Equation is shown below:

$${{Y}} = {\mathrm{Bottom}} + \frac{{{\mathrm{Top}} - {\mathrm{Bottom}}}}{{\left. {1 + 10^{({\mathrm{LogEC}}50 - {{X}})}} \right)}},$$

where X is the log of dose or concentration; Y is the Response, increasing as X increases; Top and Bottom are the Plateaus in the same units as Y, and logEC50 is the same log units as X.

To measure the Gα proteins (Gαs, Gαq, Gαi/o, Gαz) mediated activation, HEK 293T cells were co-transfected with the receptor, Gα-Rluc, Gβ3 and Gγ9-GFP in a ratio (5:1:5:5); to measure the mVenus-arrestin-1 or −2 translocation, HEK 293T cells were co-transfected with receptor-Rluc and mVenus-arrestin in a ratio (1:5). After transfection 24 h, cells were plated in the 96-well plates, and the following steps are the same as KOR-nanobody recruitment assay.

To quantify the t1/2 in Fig. 4 and Supplementary Fig. 9, data from agonist or antagonist treatment were replotted individually with a non-linear regression model using binding kinetics (Dissociation-One phase exponential decay) equation in Graphpad Prism 5. The equation “Dissociation- One phase exponential decay” is

$${{Y}} = \left( {{\mathrm{YO}} - {\mathrm{NS}}} \right) \ast {\mathrm{e}}^{ - {{K}} \ast {{X}}},$$

where X is time; Y is binding, usually total (constrain NS to 0.0 if specific); Y0 is Y at time 0, in units of Y; NS is the non-specific binding, in units of Y; K is the rate constant in inverse units of X. The half-life is 0.69/K.

Generation of chimeric receptors

To make chimeric constructs, different KOR ICL3 DNA sequence was cloned into the receptor-Rluc constructs (human serotonin 5HT2AR, 5HT4R, 5HT5AR; Dopamine D1 (DRD1); endothelin A (ETA); Neurotensin 1 (NTS1R); Histamine 2 (HRH2)). The detailed DNA and protein sequence information have been provided in the Supplementary document. For different receptors, these insertion sites may vary to obtain a maximum response. These chimeric constructs with C-terminal Rluc tagged were used in BRET assay to test Nb6 recruitment following the BRET protocol; the same chimeric constructs without C-terminal Rluc tagged were used in BRET assays to measure Gα protein activation following the protocols shown above.

To compare the kinetics between chimeric receptor/Nb6 and wild-type receptor/miniG proteins, HEK 293T cells were co-transfected with chimeric receptors with C-terminal Rluc and Nb6-mVenus (GPCR-KOR-Rluc + Nb6-mVenus) or wild-type receptor and mVenus-miniG (GPCR-Rluc + mVenus-miniG) at ratio 1:5, respectively. The transfected cells were treated following the BRET protocol. During the test, luminescence was recorded in the order: background for 30 s, + 10 µM agonist for 30 s, + 10 µM antagonist for 30 s. Cells transfected with GPCR-KOR-Rluc or GPCR-Rluc only are the background and the net BRET ratio will be the recorded value in the experimental groups subtracted by the background. Data were then plotted in Prism 5.

Radioligand binding assays

Binding assays were performed using Sf9 membrane fractions expressing the crystallization construct BRIL-KOR or HEK293 T membrane preparations transiently expressing KOR wt or KOR mutants. Binding assays were set up in 96-well plates in the standard binding buffer (50 mM Tris, 0.1 mM EDTA, 10 mM MgCl2, 0.1% BSA, pH 7.40). Saturation binding assays with 0.1–20 nM [3H]-U69,593 in standard binding buffer were performed to determine equilibrium dissociation constant (Kd) and Bmax, whereas 10 µM final concentration of JDTic was used to define nonspecific binding. For the competition binding, 50 μL each of 3H-JDTic (final 0.8 nM), drug solution (3×) and homogenous membrane solution was incubated in 96-well plate in the standard binding buffer in the absence or presence of purified nanobodies (final concentration 5 µM). Reactions (either saturation or competition binding) were incubated for 2 h at room temperature in the dark and terminated by rapid vacuum filtration onto chilled 0.3% PEI-soaked GF/A filters followed by three quick washes with cold washing buffer (50 mM Tris HCl, pH 7.40) and read. Results (with or without normalization) were analyzed using GraphPad Prism 5.0 using one site—Fit Ki model (Eqs. (3) and (4)):

$${\mathrm{logEC}}50 = {\mathrm{log}}10^{{\mathrm{logKi}}\left( {1 + \frac{{\left[ {{\mathrm{Hot}}} \right]}}{{{\mathrm{Kd}}}}} \right)}$$
$${\mathrm{Y}} = {\mathrm{Bottom}} + \frac{{{\mathrm{Top}} - {\mathrm{Bottom}}}}{{\left. {1 + 10^{({\mathrm{X}} - {\mathrm{LogEC}}50)}} \right)}}$$

where X is the log molar concentration of unlabeled ligand; Y is the binding in any convenient units.

Top and Bottom are the plateaus in units of Y-axis; logKi is the log of the equilibrium dissociation constant of tested ligand; [Hot] is the concentration of radioligand, and Kd is the equilibrium dissociation constant of radioligand.

Ligand association and dissociation assays

Binding assays were performed using membrane preparations of HEK-293T cells transiently expressing human wild-type KOR at room temperature. The same aliquots of membranes were preincubated with buffer, 5 µM Nb6 or Nb39 proteins for 30 min. Radioligand dissociation and association assays were performed in parallel using the same concentrations of radioligand, membrane preparations and binding buffer (50 mM Tris, 10 mM MgCl2, 0.1 mM EDTA, 0.1% BSA, pH 7.4). All assays used at least two concentrations of radioligand (0.6–1.5 nM[3H]-JDTic. For dissociation assays, membranes were incubated with radioligand for at least 2 h at room temperature before the addition of 10 μl of 10 μM excess cold ligand JDTic to the 200 μl membrane suspension at designated time points. For association experiments, 100 μl of radioligand was added to 100 μl membrane suspensions at designated time points.

Time points spanned 1 min to 6 h, depending on experimental conditions and radioligand. The association (Kon) and dissociation (Koff) values were calculated using the Prism model:

Association kinetics to obtain kon

$${{Y}} = {\mathrm{Ymax}} \ast (1 - {\mathrm{e}}^{ - {\mathrm{kob}} \ast {{X}}}).$$

Dissociation—one phase exponential decay to obtain koff

$${{Y}} = \left( {{\mathrm{YO}} - {\mathrm{NS}}} \right) \ast {\mathrm{e}}^{ - {{K}} \ast {{X}}},$$

X is time; Y is binding, usually total (constrain NS to 0.0 if specific); Y0 is Y at time 0, in units of Y; Ymax is the maximum binding in Y units; NS is the binding at very long times, in units of Y; K is the rate constant in inverse units of X. The half-life is 0.69/K.

To use an alternative assay to measure JDTic kinetics in the presence of Nb6 or Nb39 as shown in Supplementary Fig. 2. The kon was determined by incubation of KOR cell membranes with the indicated concentrations of [3H]-JDTic (200, 400, 800 pM) for various time periods. Assays were set up the same as the above association assay. Data were fitted using a one-phase exponential association function to yield an observed on-rate (kob) (Eq. 4). kob was plotted against the [3H]-JDTic concentration, employed for indirect determination of koff (y-intercept = 0) (Eqs. (7) and (8)):

$${\mathrm{kob}} = {\mathrm{kon}} \ast [{\mathrm{Hot}}] + {\mathrm{koff}},$$
$${\mathrm{kd}} = \frac{{{\mathrm{koff}}}}{{{\mathrm{kon}}}}.$$

For the determination of kon and koff for unlabeled Sal A, membranes containing either wild-type KOR were incubated with [3H]-JDTic and several concentrations of Sal A (1, 10, 100, 1000, 10000 nM). Non specific binding was determined by addition of 10 μM JDTic. Immediately (at time = 0 min), plates were harvested by vacuum filtration onto 0.3% polyethyleneimine pre-soaked 96-well filter mats (Perkin Elmer) using a 96-well Filtermate harvester, followed by three washes with cold wash buffer (50 mM Tris pH 7.4). Scintillation cocktail (Meltilex, Perkin Elmer) was melted onto dried filters and radioactivity was counted using a Wallac Trilux MicroBeta counter (PerkinElmer). Data were analyzed using ‘dissociation-one phase exponential decay’ or ‘association kinetics-two or more concentrations of hot radioligand’ in Graphpad Prism 5.0. The previously determined [3H]-JDTic kon and koff rates of KOR (+ buffer or Nb6 or Nb39) were used to estimate the kon and koff rates of Sal A using the ‘kinetics of competitive binding’ equation in Graphpad Prism 5.0

Kinetics of competitive binding model:

$${{Y}} = {{Q}} \ast \left( {{{K}}4 \ast \frac{{{\mathrm{DIFF}}}}{{{\mathrm{KF}} \ast {\mathrm{KS}}}} + \left( {\frac{{{{K}}4 - {\mathrm{KF}}}}{{{\mathrm{KF}}}}} \right) \ast {\mathrm{e}}^{( - {\mathrm{kF}} \ast {\mathrm{X}})} - \left( {\frac{{{{K}}4 - {\mathrm{KS}}}}{{{\mathrm{KS}}}}} \right) \ast {\mathrm{e}}^{\left( { - {\mathrm{KS}} \ast {\mathrm{X}}} \right)}} \right).$$


$$\begin{array}{*{20}{c}} {{\mathrm{KA}} = {{K}}1 \ast \left[ {\mathrm{L}} \right] \ast (1{\mathrm{e}} - 9) + {{K}}2} \\ {{\mathrm{KB}} = {{K}}3 \ast \left[ {\mathrm{L}} \right] \ast (1{\mathrm{e}} - 9) + {{K}}4} \\ {{{S}} = \sqrt {(({\mathrm{KA}} - {\mathrm{KB}})^2 + 4 \ast {{K}}1 \ast {{K}}3 \ast \left[ {\mathrm{L}} \right] \ast {\mathrm{I}} \ast (1{\mathrm{e}} - 18))} } \\ {{\mathrm{KF}} = 0.5 \ast ({\mathrm{KA}} + {\mathrm{KB}} + {{S}})} \\ {{\mathrm{KS}} = 0.5 \ast ({\mathrm{KA}} + {\mathrm{KB}} - {{S}})} \\ {{\mathrm{DIFF}} = {\mathrm{KF}} - {\mathrm{KS}}} \\ {{{Q}} = {\mathrm{Bmax}} \ast {{K}}1 \ast {\mathrm{L}} \ast \frac{{1{\mathrm{e}} - 9}}{{{\mathrm{DIFF}}}}} \end{array}$$

X is time in seconds or minutes; Y is the specific binding in cpm or some other unit.

K1 is the association rate of radioligand in inverse molar times inverse time); K2 is the dissociation constant of radioligand in inverse time), and L is the radioligand in nM). Bmax is the maximum binding, in units of Y-axis is the cold drug in nM.

K3 is the on rate of tested drug (inverse Molar times inverse time); K4 is the off rate of tested drug (inverse time);

Confocal Imaging

To prepare confocal imaging, HEK 293T cells were first plated in MatTek dishes pre-coated with poly-l-Lysine at density 75,000 cells/plate in 10% FBS DMEM. The next day KOR-mScarlet and Nb6-mVenus or Nb39-mVenus were co-transfected with TransIT-2020 transfection reagent. These reporters were co-expressed in HEK 293T cells with a 4:1 ratio to ensure an excess of receptor and minimize free nanobody. For Golgi labeling, KOR, Nb39-mVenus and GalT-RFP were co-transfected. After at least 16 h, the medium was changed to 1% dFBS DMEM and incubated at 37 for another 24 h. On the experiment day, cell medium was removed and changed to drug buffer (20 mM HEPES, 1X HBSS, pH 7.4 supplemented with 0.1% BSA, 0.01% L-ascorbic acid and 4.5 g/L Glucose). Measurements were carried out on an inverted Olympus FV3000RS confocal microscope equipped with a Tokai Hit WSKMX live-cell chamber (set to maintain sample heat at room temperature) and a LUMPL FL 60× SuperCorr Plan ApoN 1.40 NA oil immersion objective and four cooled GaAsP photomultiplier detectors. One-way/Round trip galvanometer/resonant scanning was used for acquisition with a 1× optical zoom. The 488 nm diode laser was used for the mVenus excitation and the 561 nm diode laser was used to excite the mScarlet or RFP, with the laser power set between 0.1 and 5% according to the power slider in the FV31S-SW microscope software, for monitoring fluorescence intensities. The variable slit had a nominal bandwidth of 500–545 nm for the green emission channel and 590–650 nm for the red emission channel.

All image processing was done using open-source Fiji54. To quantify KOR–Nb6 translocation, Binary Cell masks (having values of 255 or 0) were created by thresholding Nb6-mVenus intensity in single-cell images at each time point, using Triangle segmentation and filling all holes were using binary masking function. A set of Cell ring masks were created by successive erosion of mask images by 15 pixels and subtracting inner region masks. One-bit masks (having values of 1 or 0) were multiplied with the original cell images to create 16-bit images for the outer and inner rings. The Ratio of the inner Median pixel (cytosol) value to the outer ring (membrane) pixel value was measured and plotted as a function of time.

To quantify KOR–Nb39 translocation, the Co-localization of Golgi was measured by masking images for bright signals in green (Nb39-mVenus) and red (GalT-RFP) channels. Golgi masks were created by creating masks representing the overlap of these red and green binary masks using the Boolean AND function. The Golgi masks were multiplied with the original NB39-mVenus cell images for each time point, measured and normalized to the peak values.

cAMP inhibition assay

To measure KOR/Gαi-mediated cAMP inhibition, HEK 293T (ATCC CRL-11268) cells were co-transfected with human KOR along with a luciferase-based cAMP biosensor (GloSensor; Promega) and with or without varying concentration of Nb6 DNA, and assays were performed similarly to previously described55. After 16 h, transfected cells were plated into Poly-lysine coated 384-well white clear bottom cell culture plates with DMEM + 1% dialyzed FBS at a density of 15,000–20,000 cells per 40 μl per well and incubated at 37 °C with 5% CO2 overnight. The next day, drug solutions were prepared in fresh drug buffer [20 mM HEPES, 1X HBSS, 0.3% bovine serum album (BSA), pH 7.4] at 3X drug concentration. Plates were decanted and received 20 μl per well of drug buffer (20 mM HEPES, 1X HBSS) followed by the addition of 10 μl of drug solution (3 wells per condition) for 15 min in the dark at room temperature. To stimulate endogenous cAMP via β adrenergic-Gs activation, 10 μl luciferin (4 mM final concentration) supplemented with isoproterenol (400 nM final concentration) were added per well. Cells were again incubated in the dark at room temperature for 15 min, and luminescence intensity was quantified using a Wallac TriLux microbeta (Perkin Elmer) luminescence counter. Results (relative luminescence units) were plotted as a function of drug concentration, normalized to % SalA stimulation, and analyzed using log(agonist) vs. response in GraphPad Prism 5.0 (Eq. 1).

Tango arrestin recruitment assay

The KOR Tango constructs were designed and assays were performed as previously described56,57. HTLA cells expressing TEV fused-β-Arrestin2 (kindly provided by Dr. Richard Axel, Columbia Univ.) were transfected with the KOR Tango construct with or without varying concentration of Nb6 DNA. The next day, cells were plated in DMEM supplemented with 1% dialyzed FBS in poly-l-lysine coated 384-well white clear bottom cell culture plates at a density of 10,000–15,000 cells/well in a total of 40 µl. The cells were incubated for at least 6 h before receiving drug stimulation. Drug solutions were prepared in drug buffer (20 mM HEPES, 1X HBSS, 0.3% BSA, pH 7.4) at 3X and added to cells (20 µl per well) for overnight incubation. Drug solutions used for the Tango assay were exactly the same as used for the cAMP assay. The next day, media and drug solutions were removed and 20 µl per well of BrightGlo reagent (purchased from Promega, after 1:20 dilution) was added. The plate was incubated for 20 min at room temperature in the dark before being counted using a luminescence counter. Results (relative luminescence units) were plotted as a function of drug concentration, normalized to % SalA stimulation, and analyzed using log(agonist) vs. response in GraphPad Prism 5.0 (Eq. 1).

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.