Orchestrating NTSR1 signaling from the interface

Biased allosteric modulators provide great therapeutic potential by selectively directing signal bias in the presence of endogenous ligand under (patho)physiological conditions. In a recent Cell Research paper, Sun et al. revealed the structural mechanisms underlying the biased allosteric modulation exerted by SBI-533 directly at the neurotensin receptor 1–β-arrestin1 interface.

Upon activation, G protein-coupled receptors (GPCRs) elicit diverse downstream signaling cascades by coupling to different heterotrimeric G proteins. These signals are spatially and temporally regulated through the coupling to the non-G protein effectors such as GPCR protein kinases (GRKs) and arrestins. Biased signaling is a phenomenon in which GPCR is stabilized in a certain conformation that preferentially activates certain pathways over the others.¹ This concept has attracted significant attention in GPCR drug discovery due to the potential for selectively activating the beneficial signaling pathways while minimizing activation of those associated with side effects. Acting on less-conserved allosteric sites, allosteric modulators show advantages in finely tuning the actions of orthosteric ligands selectively. Biased allosteric modulators (BAMs) therefore represent a promising class of GPCR ligands for developing safer and more effective therapeutics.²

Through high-throughput screening and medicinal chemistry optimization, SBI-553 was discovered as a small-molecule BAM that selectively modulates neurotensin receptor 1 (NTSR1). It enhances NTSR1-mediated β-arrestin1/2 (βArr1/2) recruitment while hindering the activation of G_q protein in the presence of neurotensin (NTS).^3,4,5 Given its high oral bioavailability, brain-penetrant property and interesting BAM activity, SBI-553 represents a promising candidate for developing safe therapeutics targeting NTSR1 for anti-psychostimulant effects.^3,4,5 It is therefore crucial to understand the molecular mechanisms of this pathway-selective allosteric modulation. Using cryo-electron microscopy (cryo-EM), previous studies have elucidated the structures of NTS–NTSR1 in complex with GRK2⁶/G_o protein⁵ with and without SBI-553 bound, as well as the NTS–NTSR1–βArr complexes.^7,8 In a recent study, Sun et al.⁹ resolved the final piece of the puzzle by elucidating the structure of the full-length NTSR1 with βArr1 bound in the presence of both SBI-553 and NTS, revealing a new engagement configuration of the GPCR–βArr1 complex.

To determine a high-resolution structure, highly stable and homogenous complex samples are essential. Sun et al. employed a chemical protein synthesis strategy to obtain full-length NTSR1 with six phosphorylated residues at its C-terminus for βArr1 binding. They first identified and validated the phosphorylation sites essential for high-affinity βArr1 binding via fluorescence polarization assay. Next, the authors synthesized the hexa-phosphorylated C-terminal tail (C-tail) of NTSR1 with microwave-assisted solid phase peptide synthesis. This phosphorylated C-tail and NTSR1 transmembrane domain were fused with the segments from the split Cfa DnaE intein. The engineered split intein segments directed both parts of the NTSR1 to join, followed by trans-splicing reaction removing the intein itself, leading to the generation of full-length phosphorylated NTSR1. This method ensured a uniform and precisely controlled phosphorylation pattern of GPCR, compared to the traditional way which relies on the action of GRK.

The authors then assembled this phosphorylated NTSR1 with constitutively active βArr1 and NTS_8–13, with or without SBI-553, for structural determination using cryo-EM. A fragment antigen-binding region Fab30 and phosphatidylinositol-4,5-bisphosphate were also added to further enhance the stability of the complexes. With these optimizations, Sun et al. successfully determined the cryo-EM structure of the active NTSR1–βArr1 complex at a higher resolution than all the previously reported GPCR–βArr1 structures.

Consistent with the previously reported NTSR1–βArr1 structures,^7,8 NTSR1 interacts with βArr1 through both ‘core engagement’ and ‘tail engagement’, where its intracellular transmembrane cavity interacts with the central crest of βArr1 and its phosphorylated C-tail engages with the N-terminal lobe of βΑrr1. These two primary interfaces were found in the complexes regardless of the presence of SBI-553. Although the binding of SBI-553 in the intracellular cavity slightly pushes down the finger loop in the central crest of the βArr1, the interaction between NTSR1 and βΑrr1 is not disrupted. SBI-553 directly interacts with the hydrophobic intracellular cavity of NTSR1 via its pendant phenyl ring and engages the finger loop of βArr1 through its quinazoline group, bridging and stabilizing the interactions between NTSR1 and βArr1.

In addition to the conventional core engagement configuration, Sun and colleagues discovered a new ‘loop engagement’ configuration that is only observed in the SBI-553-bound complex but not in the SBI-553-unbound complex (Fig. 1a). In this ‘loop engagement’ complex, both intracellular loop 1 (ICL1) and ICL3 of NTSR1 are more extended. The quinazoline group interacting with the finger loop of βArr1 in the conventional ‘core engagement’, now forms hydrophobic interactions with the ICL3 of NTSR1. This, in turn, bent the loop towards the intracellular cavity and stabilized it, which enables the ICL3 to interact with the finger, middle, gate and lariat loops in the central crest of βArr1. In addition, the more extended ICL1 of the NTSR1 forms strong interactions with the lariat loop of βArr1. The interactions between these key residues were confirmed using mutagenesis study. Other than these additional interfaces, Sun et al. found that the engagement of the phosphorylated C-tail of NTSR1 with βArr1 is still maintained. Using PDBePISA, the authors quantified the total interaction area, revealing that the ‘loop engagement’ configuration exhibits stronger interactions between NTSR1 and βArr1 than the ‘core engagement’ configuration. Importantly, Sun et al. suggested that the orientation of these ICLs stabilized by SBI-553 only fits well with the loops in βArr1 but creates steric hinderance for G protein binding based on the overlayed structural models (Fig. 1b).

Fig. 1: Structural mechanisms of the biased allosteric modulation exerted by SBI-553 at NTSR1.

a, b BAM SBI-553 potentiates βArr1 recruitment (a) and prevents G_q protein recruitment (b) by binding to the intracellular cavity of active NTSR1. a Structural model of NTSR1–βArr1 complexes in the absence of SBI-553 (top, PDB: 6UP7⁸) and in the presence of SBI-553 (bottom left, PDB: 8ZYU; bottom right, PDB: 8ZYT⁹). SBI-533 could either directly bind to the finger loop of βΑrr1 to facilitate the interactions (core engagement) or orientate the ICL3 of NTSR1 to form extensive interactions with the loops in βArr1 (loop engagement). The bottom part of NTSR1 is shown in light gray cartoon with the loops interacting with βArr1 labeled in dark gray. βArr1 in SBI533-unbound/core engagement/loop engagement configuration is labeled in pink/blue/green cartoon, respectively, with the key interacting loops highlighted. Critical residues interacting with SBI-553 are shown in sticks. b Structural model of the active NTSR1–G_q complex (top, PDB: 8FMZ⁵). Structural overlay of the active NTSR1–G_q complex with the SBI-553–NTSR1 from the SBI-553–NTSR1–βΑrr1 core engagement complex (bottom left) or from the SBI-553–NTSR1–βΑrr1 loop engagement complex (bottom right). The structural overlay suggests that SBI-533 occupies the binding site of the C-terminus of G_q α5 helix (orange helix). In addition, active NTSR1 with ICL3 bent and stabilized by SBI-553 could not bind to G_q protein due to strong steric clashes.

In summary, both SBI-553-bound NTSR1 configurations favor βArr1 binding over the other transducers, explaining the BAM activity of SBI-553 from an elegant structural perspective. Superimposition of the arrestin complex structures with previously reported structures of GRK/G_oA complexes reveals a strikingly conserved NTSR1 receptor core with only changes in ICL1 and ICL3 as discussed earlier. Additionally, the pose of SBI-553 quinazoline moiety is also well conserved in these structures with only minor displacement of the pendant phenyl moiety. The flexibility of its binding pose within the hydrophobic cavity provides great potential for future drug discovery. Together, these structural insights could further guide the design and development of BAMs, for precisely tuning biased signaling directly from the receptor–transducer interface.