Somatostatin (SST), an inhibitory hormone that distributes widely in both human central nervous system and periphery tissues negatively regulates multiple hormone releases (growth hormone, glucagon, insulin, gastrin and cholecystokinin) and cell proliferation through SST receptors (SSTRs). There are five types of SSTRs, namely SSTR1–SSTR5, which are divided into two subfamilies: SRIF1 (SSTR2, SSTR3, SSTR5) and SRIF2 (SSTR1, SSTR4) based on phylogenetic, sequence homology and ligand binding profiles. Two bioactive SST peptides, SST-14 and its N-terminally extended form SST-28, have been identified in mammals and show high and equal affinity for SSTR1–5. Among all the SSTRs, SSTR2 is the best characterized member with multiple effects on hormone secretion, cell cycling, apoptosis and angiogenesis. It is also the most common subtype expressed in both human neuroendocrine tumors (NETs) and related hormone diseases, making it a valuable target for diagnosis and therapy of tumors as well as acromegaly. On the contrary, SSTR4 is highly expressed in the central nervous system and mediates potent analgesic and anti-inflammatory actions. In addition, studies in recent years have revealed that SSTR4 agonists represent a promise for non-opioid pain control, especially for chronic neuropathic, inflammatory and mixed pain.
Along with increased knowledge of the pharmacological effects of these receptors, medical application of SST and its analogs has expanded. A considerable amount of effort has been made to develop new therapeutics for oncology of SSTRs. Due to a very short half-life of SST (less than 3 min), several of its agonist analogs, especially octreotide and lanreotide, are used to treat acromegaly and NETs by targeting SSTR2, while its antagonist analog, CYN 154806, has been employed to study diverse functions of this receptor. It was found recently that radiolabeled SSTR antagonists produced superior images than that of agonists, pointing to a potential application in imaging and treating SSTR-expressing tumors. However, there are concerns about limited effectiveness and adverse events such as gastrointestinal disturbance and hyperglycemia. Since peptides have relatively short half-life and poor penetration to the blood–brain barrier, non-peptide ligands with high potency and subtype selectivity have been developed for each SSTR subtype showing different pharmacological properties. L-054,522 was identified and optimized by Merck that mimics the side chains of W8 and K9 at the β turn tip of the endogenous peptide SST-14, and displayed at least 3000-fold better selectivity for SSTR2. This full agonist exerts an inhibitory effect on growth hormone and glucagon releases, while SSTR4 selective agonist, J-2156, mediates pain relief.
To reveal ligand selectivity and activation mechanisms of SSTRs, we solved the crystal structures of SSTR2 bound to selective peptide antagonist CYN 154806 and non-peptide agonist L-054,522, as well as cryo-EM complex structures of SSTR2–G i1 bound to endogenous ligand SST-14, SSTR4–G i1 bound to SST-14 and SSTR4–G i1 bound to non-peptide agonist J-2156, respectively. Combined with mutagenesis, molecular docking and molecular dynamics (MD) simulation studies, these structures reveal the key signature shared by their ligands which is prerequisite for receptor binding. Our findings also provide molecular insights into ligand selectivity, receptor activation and G protein coupling thereby offering near-atomic-resolution models for rational design of better drugs against SSTRs.
To facilitate the lipidic cubic phase (LCP)crystallization of SSTR2 with its peptide antagonist CYN 154806, the flexible C-terminus was truncated to T359 and bacillus subtilis xylanase was inserted between S238 and G243 of the intracellular loop 3 (ICL3). Three mutations, D89 2.50 N, V106 ECL1 E and S316 8.47 D (superscript numbers represent Ballesteros–Weinstein nomenclature), were introduced to improve protein yield and homogeneity. To solve the crystal structure of L-054,522-bound SSTR2, D89 2.50 was reinstated as in wild type (WT) and the junction site of xylanase was adjusted between I240 and V242 to improve crystal quality. Crystals of both complexes were obtained in monoolein lipid phases and determined at 2.65 Å and 2.6 Å resolution, respectively (Supplementary information, Table S1). These modifications had decreased the ligand binding for CYN 154806 by ~50-fold compared with the WT but had little effect on L-054,522 binding and signaling (Supplementary information, Fig. S1a, b and Tables S3 and S4).
For cryo-EM studies, SSTR2 WT with C-terminal truncation to T359 and SSTR4 with C-terminal truncation to L328 plus a V264 6.40 F mutation were prepared to facilitate complex formation. Binding and signaling assays showed that these modifications had no influence on receptor activities (Supplementary information, Fig. S1a–d and Tables S3 and S4). To obtain stable complexes, three subunits of G i1 protein were co-expressed with the receptors in High-Five insect cells. Complexes were assembled in the membrane and a single-chain variable Fab fragment (scFv16) was applied to stabilize the SSTR4–G i1 complexes. Structures were determined by single-particle cryo-EM at a nominal resolution of 3.1 Å (SST-14–SSTR2–G i1), 2.9 Å (SST-14–SSTR4–G i1) and 2.8 Å (J-2156–SSTR4–G i1), respectively (Supplementary information, Figs. S2, S3 and Table S2).
The overall structures of SSTR2 and SSTR4 possess the canonical seven-transmembrane (7-TM) architecture with an extended helix VIII in parallel with the membrane (Fig. 1a, b). Similar to other solved peptide-bound class A G protein-coupled receptors (GPCRs), extracellular loop 2 (ECL2) of SSTR2 and SSTR4 form short antiparallel β-strands stabilized by conserved disulfide bonds between Cys 3.25 and Cys ECL2 (Fig. 1a, b). Among the five structures, two crystal structures are in inactive state or in agonist-bound inactive state, probably due to the fact that conformations with low energy states facilitated the crystallization (Fig. 1c). In contrast, the G i1-coupled SSTR2 and SSTR4 adopt full active states with a remarkable outward displacement of helix VI (~10 Å, measured by Cα of 6.29) accompanied by transverse movement of helix V and inward movement of helix VII, which are consistent with activation characteristics of class A GPCRs (Fig. 1c). In spite of binding to G i1 and different ligands, SSTR2 and SSTR4 structurally resemble each other with root-mean-square-deviation (RMSD) values of 1.3–1.5 Å for the Cα atoms (Supplementary information, Fig. S4a). However, G protein binding between the two receptors still exhibit several conformational differences. Compared to the SST-14–SSTR4–G i1, the SST-14–SSTR2–G i1 complex shows that the C-terminus of α5 helix of Gα i1 tilts ~2 Å toward helix VI (measured by the Cα of F354 of Gα i1), which further induces an outward movement of helix VI but inward movement of ICL3 of SSTR2 as opposed to SSTR4 (Supplementary information, Fig. S4b). It appears that diverse residues of ICL3 between SSTR2 and SSTR4 form different interactions with Gα i1. S244 ICL3 of SSTR2 makes polar interactions with the side chain of E318 of Gα i1; however, W247 ICL3, its counterpart in SSTR4, pushes the ICL3 away due to a bulky side chain (Supplementary information, Fig. S4c). The ICL3 in closer proximity forms further interactions between SSTR2 and Gα i1, e.g., K246 ICL3 makes a hydrogen bond with the main chain of D315 of Gα i1, which is not observed in SSTR4 (Supplementary information, Fig. S4c). The closer contacts lead to a larger interaction area between SSTR2 and Gα i1 (~1129 Å 2) than that between SSTR4 and Gα i1 (~965 Å 2), thereby contributing to a stronger binding of Gα i1 towards SSTR2. Indeed, according to our binding and signaling data, even though SST-14 binds to both receptors with similar affinities (1.4 nM vs 1.5 nM), it displays 5-fold higher potency in activating SSTR2 compared to SSTR4 (Supplementary information
