2.1 Cryo-EM structures of Y₁R/Y₂R–Gi2 signaling complexes

In order to investigate Y₁R and Y₂R activation potency in response to NPY variants, including N-terminally truncated NPY and NPY mutants (Figure 1A), we measured the cAMP inhibition ability of wildtype Y₁R and Y₂R elicited by different agonists in HEK293 cells (Figures 1A–C). Our results indicated that deleting the N-terminal residues (Y¹ and P²) of NPY (NPY³⁻³⁶) and NPY¹³⁻³⁶ remarkably reduced Y₁R activation compared to NPY (>50 folds), whereas [Leu³¹, Pro³⁴]–NPY analog retained a potency nearly identical to NPY (Figure 1B). In contrast, the NPY³⁻³⁶ variant exhibited slightly reduced Y₂R activation and [Leu³¹, Pro³⁴]–NPY was ∼100-fold less potent than NPY on Y₂R activation (Figure 1C). These observations suggest that Y₁R and Y₂R display different ligand recognition mechanisms.

To gain a more thorough understanding of how Y₁R and Y₂R recognize NPY or NPY mutants, we determined the structures of NPY-bound Y₁R, NPY-bound Y₂R, and [Leu³¹, Pro³⁴]–NPY-bound Y₁R in complex with the Gi2 protein using cryo-EM single-particle technique (Figures 1D–I and S1 and 2, and Table S1). The final cryo-EM maps of NPY–Y₁R–Gi2 (PDB ID:8K6M), NPY–Y₂R–Gi2 (PDB ID:8K6N), and [Leu³¹, Pro³⁴]–NPY–Y₁R–Gi2 (PDB ID: 8K6O) displayed global nominal resolutions of 3.5, 3.2, and 3.3 Å after refinement, respectively (Figures S1 and 2 and Table S1). The Y₁R and Y₂R structure shared a canonical seven-transmembrane (TM) helical architecture and an intracellular amphipathic helix 8. The seven TM helix domains of both receptors could be visibly distinguished in the cryo-EM maps (Figure S3), and the high-resolution density maps allow us to unambiguously build the NPY, [Leu³¹, Pro³⁴]–NPY, NPYRs, Gαi₂, Gβ, Gγ, and scFv16 (Figures 1D–I and S3). However, probably due to the NPY flexibility in the extracellular region, we failed to model the residues 10−15 from NPY and [Leu³¹, Pro³⁴]–NPY and the residues 10−14 from NPY in Y₁R and Y₂R complex structures, respectively (Figures 1G–I).

By comparing NPY–Y₁R or NPY–Y₂R with their inactive states (PDB ID: 5ZBQ and 7VGX, respectively),^{31, 32} we noticed that the intracellular end of TM6 in two receptors exhibited an obvious outward shift to accommodate the Gαi₂ α5 helix and TM7 also adopted an inward displacement similar to other active class A GPCR structures^33-35 (Figures S4A and B). As the NPY–Y₁R and [Leu³¹, Pro³⁴]–NPY–Y₁R structures share highly similar conformation with a Cα root mean square deviation (RMSD) of 0.569 Å (279–279 atoms), we use the NPY–Y₁R–Gi2 and NPY–Y₂R–Gi2 structures to analyze the overall Y₁R and Y₂R structures in this section (Figure S4C).

Although Y₁R and Y₂R shared only 30% sequence identity, these two receptors activation complex structures resembled a similar conformation with a Cα RMSD of 0.905 Å (213–213 atoms) (Figure S4D). It is noteworthy that the extracellular end of TM1, TM4, and TM5 in Y₁R exhibited a slight outward movement relative to that in Y₂R (Figure S4D). The ligand NPY displayed similar binding pattern both in the Y₁R and Y₂R structures and folded into a canonical PP-fold,³⁶ the C-terminal regions (residues 32−36) were inserted into an orthosteric pocket of Y₁R and Y₂R, while the N-terminals (residues 1−10) folded back to form extensive interactions with the NPY α-helix (residues 15−31) (Figures 1G–H and S5A–C).

2.2 A sub-pocket in Y₁R determines NPY selectivity

The NPY–Y₁R with NPY–Y₂R structural comparison revealed remarkable displacements (∼9 Å) in the NPY N-terminal part (Figure S5D). In Y₁R, the NPY N-terminal end could be found in a cavity composed of the extracellular ends of TM5, TM6, and ECL2. However, the NPY N-terminal part in Y₂R extended toward the ECL2 region, forming relatively few contacts with the receptor. In accordance with the NPY pharmacological assay, the structure revealed different NPY recognition mechanisms in Y₁R and Y₂R.

In detail, the NPY N-terminal Y¹ and P² residues fitted into the amphiphilic cavity formed by hydrophobic (P183^ECL2, F199^ECL2, and F286^6.58) and polar (E182^ECL2 and D200^ECL2) residues, and the corresponding [Leu³¹, Pro³⁴]–NPY residue adopted a conformation similar to that of NPY (Figure S6A). Therefore, we defined the cavity in Y₁R as a unique sub-pocket (Figures 2A and B), where the clusters of hydrophobic residues (P183^ECL2, F199^ECL2, and F286^6.58) formed a hydrophobic contact network with Y¹ and P² residues and D200^ECL2 and E182^ECL2 established a polar network with the NPY peptide Y¹ and K⁴ side chains (Figure 2B). In addition, two critical residues R208^6.35 and F286^6.58 bifurcated the NPY binding pocket in Y₁R into two cavities, a sub-pocket, and a major orthosteric pocket (Figure 2B). In contrast to the Y₁R NPY recognition mode, Y₂R did not contain the equivalent sub-pocket for NPY binding (Figures 2C and D).

To further identify the key residues of the Y₁R sub-pocket for the NPY N-terminus recognition, we conducted a series of mutagenesis analyses. The Y₁R and Y₂R sequence alignment results confirmed that the Y₁R sub-pocket residues (E182^ECL2, D200^ECL2, P183^ECL2, F199^ECL2, and F286^6.58) were not conserved (Figure 2E). We replaced these key residues with the corresponding residues in Y₂R and our mutagenesis test demonstrated that mutations E182I, D205S, and H290S in Y₁R did not affect the receptor activation potency (Figure 2F and Table S2). However, the hydrophobic residues of sub-pocket mutations (P183^ECL2E, F199^ECL2T, F286^6.58 V, and P183^ECL2E/F199 ^ECL2T/F286^6.58 V) all severely reduced Y₁R potency to NPY (Figure 2G and Table S2). Our results indicate that the sub-pocket-forming hydrophobic residues contribute to the Y₁R-mediated NPY recognition.

Compared with other β-branch structures of class A GPCRs, for instance, the endogenous endothelin ET-1 peptide-bound ET_B receptor structure revealed a similar peptide–receptor binding mechanism as that in Y₁R, in which the extracellular part of the receptor (ECL2, TM6, and ECL3) is widely involved in the N-terminal ET-1 binding³³ (Figure S6B). However, the N-terminal ET-1 region mainly forms hydrogen bonds and electrostatic interactions with ET-1 and no such Y₁R sub-pocket has been observed in the ET_B–ET1 structure (Figures S6B–C).

Our structures revealed a sub-pocket, which plays important roles in the N-terminal NPY recognition and receptor activation of Y₁R. Our observation was consistent with previous Y₁R-related NPY NMR modeling and docking studies.³⁷ Taken together, our study provides insights into the detailed NPY N-terminus-Y₁R interactions and helps in fully understanding the different selectivity mechanisms of various N-terminally truncated NPY peptides upon Y₁R and Y₂R activation.

2.3 Adaptive evolution of the Y₁R sub-pocket

NPY and Y₁R evolutionary dynamics in early vertebrates have already been thoroughly studied.^{2, 38} However, the evolutionary fate of Y₁R sub-pocket is elusive. We collected 22 genomes of vertebrates and invertebrates to reveal the related evolutionary processes. Our results revealed that NPY and Y₁R were present in vertebrates such as Cyclostomata, Chondrichthyes, Osteichthyes, and Tetrapoda, but not in invertebrates (Figure 3A). We selected six classical vertebrate Y1Rs for signal pathway detection and the results demonstrated that Cyclostomata (pmY₁R/Petromyzon marinus), Osteichthyes (drY₁R/Danio rerio), and Tetrapoda (hY₁R/Homo Sapiens, and cpbY₁R/Chrysemys picta bellii) Y1Rs could be activated by NPY, while such activation is significantly weakened (scY₁R EC₅₀ = 175.1 nM and ccY₁R could not be activated by NPY) in Chondrichthyes (scY₁R/Scyliorhinus canicula and ccY₁R/Carcharodon carcharias) (Figures 3A and B). The branch-site model has proved to be a useful tool for detecting biological hypotheses of positive selection and generative mutation research and functional analysis.³⁹ So, we used a branch-site model to assess whether the Y₁Rs underwent positive selection.^{40, 41} As shown in Figure 3C, the Y₁R sequences exhibited a relevant nonsynonymous (dN)/synonymous (dS) substitution rate ratio (branch-site dN/dS of ω > 1; Figures 3A and C) that was highly significant (likelihood ratio tests [LRTs], p < 0.05; Figure 3C) only in the Osteichthyes lineage. In contrast, Chondrichthyes did not exhibit this ratio, suggesting that positive Darwinian selection occurred specifically in Osteichthyes, but not in Chondrichthyes Y₁Rs (Figure 3C). And we found that the sub-pocket residues Q182 and R203 in Osteichthyes Y₁R were under positive selection (Figure 3C). Furthermore, according to sequence logos between Osteichthyes and Chondrichthyes Y₁Rs, the six sub-pocket residues were non-conserved (Figure 3D).

To verify the Y₁R-related functional differences between Chondrichthyes and other vertebrates, three vertebrate species (human/Homo Sapiens, zebrafish/Danio rerio, and Carcharodon/Carcharodon carcharias representing Tetrapoda, Osteichthyes, and Chondrichthyes, respectively) were selected to test the biological experimental research (Figure 3E). We used the NPY peptide to activate the different Y₁Rs. NPY activated humans and zebrafish, but not Carcharodon, Y₁Rs. To further investigate the function of the six residues (E182^ECL2, Q186 ^ECL2, F199 ^ECL2, D200^ECL2, D205^5.32, F286^6.58, and H290^ECL3) in the sub-pocket, the human Y₁R residues (hY₁R) were mutated to the corresponding residues in the Chondrichthyes orthologs. Similar to the effect of mutating the corresponding key residues in Y₂R, residue F199^ECL2 and F286^6.58 substitutions markedly affected the hY₁R activation efficacy in response to the NPY peptide (Figure 3E). In addition, residues N186 and D205 in human Y₁R sub-pocket under positive selection were also engaged in NPY recognition (Figure S7A). Next, Carcharodon Y₁R (ccY₁R) (Q181^ECL2, Y184 ^ECL2, I199 ^ECL2, E205^5.32, and Y290^ECL3) and zebrafish Y₁R (drY₁R) (A188 ^ECL2, Q192 ^ECL2, V208 ^ECL2, R214^5.32, and I295^6.58) residues were reversely mutated to the corresponding hY₁R. We observed that mutated residues in Carcharodon Y₁R to the corresponding hY₁R does not rescue their ability to respond to the NPY. From an evolutionary perspective, the reason is that the protein conformation of Carcharodon Y₁R has undergone significant changes, and rescuing their ability to respond to NPY may require more than a few point mutations (Figures 3E and S7B-C). Taken together, our results indicated that the positive selection pressure in Osteichthyes Y₁R maintains the sub-pocket and contributes to retaining the Gi signaling. In contrast, Chondrichthyes were not constrained by the selection pressure, and the mutations in the sub-pocket amino acids led to changes in their Gi signaling.

2.4 Conserved common site for NPY recognition by NPYRs

Overall, the NPY-bound Y₁R and Y₂R structures shared a similar conformation and central residues, E15–L31 (α-helix region) of NPY formed approximately five α-helical turns and the base of the NPY α-helix overlay in the two structures, while the amino terminus of the Y₁R helix was rotated approximately 10˚ inwards compared with that in Y₂R (Figure S8A). The residue R²⁵ of NPY in the Y₁R structure formed hydrogen bonds with D104^2.68, which could not be observed in the NPY-Y₂R structure, potentially contributing to the inwards shift of the NPY-bound Y₁R α-helix (Figure S8B). In addition, the hydrogen bonds between the residue R²⁵ of NPY and D104^2.68 further stabilized the NPY–Y₁R interactions. Consistent with these structural observations, a D104A^2.68 substitution reduced the potency of the Y₁R response for NPY (∼73 folds) (Figure S8C).

NPY in the Y₁R and Y₂R complex structures adopted similar conformations in the recognition by the NPYR orthosteric pockets (Figures 4A–D). The C-terminus of NPY (residues 33−36), which was confirmed to display major importance for all NPYR bindings,^{3, 42} adopted an extended conformation and reaches far into the cores of Y₁R and Y₂R, contacting all TM helices except for TM1, as well as residues in ECL2 and ECL3 through an extensive interface of hydrophobic and polar interactions (Figures 4A–D). The NPY C-terminal-bound Y₁R structure exhibited both important common characteristics and notably distinct features compared to the NPY-bound Y₂R structure (Figures 4A–D). The structural comparison revealed that the Y₁R-bound NPY side chains R³⁵ and Y³⁶ overlaid well with those of Y₂R-bound NPY, while the NPY residues R³³ and Q³⁴ occupied different binding sites to those of Y₂R-bound NPY (Figure 4E).

At the bottom region of the orthosteric peptide-binding pocket, polar receptor residues formed an extensive polar interaction network with amidated Y³⁶ both in the Y₁R-and Y₂R-bound NPY structures, structurally supporting the fact that this amidation modification was necessary for the NPYR activity.^{43, 44} The Y³⁶ amidation group and hydrogen bond established polar contacts with residues T107^2.61 and S223^5.46 in the NPY-Y₂R structure (Figures 4C and D), whereas the amidation and carbonyl groups of Y³⁶in the Y₁R-bound NPY formed hydrogen bonds with residues Q120^3.32 and H306^7.39, respectively (Figures 4A and B). In addition, residue Y³⁶ was further coordinated by Q219^5.46 through polar interactions and C93^2.57 through van der Waals contacts in the NPY-Y₁R structure (Figures 4A and B). On the same Y³⁶ orientation, residue R³⁵ was fastened mainly through polar interactions by D^6.59 in the NPY–Y₁R and NPY–Y₂R structures (Figures 4A–D). Notably, previous studies reported that the ionic interaction of residue D^6.59 is key for positively charged ligand recognition.⁴⁵

To further identify the common binding sites NPYRs, we conducted a series of mutagenesis analyses. Our NPYR sequence alignments indicated that residues (C^2.57, T^2.61, Q^3.32, and D^6.59) in TM2, TM3, and TM6, which form polar interactions and van der Waals contacts with the C-terminus residues Y³⁶ and R³⁵, were conserved among the four receptor subtypes (Figures 4E and F). Mutation of these conserved residues (C^2.57A, T^2.61A, Q^3.32A, and D^6.59A) in Y₁R, Y₂R, Y₄R, and Y₅R significantly impaired receptor activity (Figures 4G, S9–12 and Table S2), supporting the essential role of these residues in NPY recognition. Taken together, the structural observations and mutagenesis analyses revealed that NPYRs share common binding sites to bind NPY residues Y³⁶ and R³⁵ and adopt different molecular patterns in the interaction with NPY.

2.5 The different recognition patterns in Y₁R and Y₂R

Our cAMP assay data demonstrated [Leu³¹, Pro³⁴]–NPY retained similar activation potency as NPY on Y₁R, but displayed higher selectivity for Y₁R over Y₂R, indicating that both Y₁R and Y₂R has different mechanisms for ligand recognition. Our structures of Y₁R and Y₂R in complex with NPY or [Leu³¹, Pro³⁴]–NPY offer templates to understand the ligand recognition or selectivity. In the case of the NPY-Y₁R structure, the residue Q³⁴ is observed to interact with only T97^2.61 in Y₁R through hydrogen bonding (Figure 5A). Whereas the side chain of Q³⁴ in Y₂R displayed a different conformation and was projected into a cavity shaped by TM2 and TM7, interacting with T107^2.61, Y110^2.64, T111^2.65, and F307^7.35 in Y₂R (Figures 5B and S13A).

To further analyze the selectivity of [Leu³¹, Pro³⁴]–NPY, we determined the structure of the [Leu³¹, Pro³⁴]–NPY-bound Y₁R in complex with the Gi2 protein. The C-terminal parts of NPY and [Leu³¹, Pro³⁴]–NPY adopted similar binding pose in the orthosteric pocket of Y₁R (Figure S13B). Residue P³⁴ of [Leu³¹, Pro³⁴]–NPY share a similar position with residue Q³⁴ of NPY (Figures 5C and S13B and C). The substitution of Q with P at position 34 of NPY disrupted the interactions between Q34 and T97^2.61 in Y₁R. The residue P³⁴ in [Leu³¹, Pro³⁴]–NPY peptide was not observed to make polar interaction with Y₁R, and this kind of peptide did not influence Y₁R signaling activation. Comparing to structures of Y₁R bound NPY or [Leu³¹, Pro³⁴]–NPY, the residue Q³⁴ of NPY makes hydrogen bonds with T107^2.61 and T111^2.65 in Y₂R (Figure 5C). These residues T107^2.61, Y110^2.64, T111^2.65, and F307^7.35 that involved in ligand binding pocket are conserved in both Y₁R and Y₂R. We next investigated whether these conserved residues play different roles in Y₁R and Y₂R, the residues T^2.61, Y^2.64, T^2.65, and F^7.35 were replaced with Ala both in Y₁R or Y₂R. The results of our functional assays showed that all mutants of Y₂R significant lost NPY-induced signal transduction, which further confirm the important roles of these residues in Y₂R (Figure S13D). By contrast, T^2.65A and F^7.35A mutations in Y₁R displayed similar activation potency as wild-type receptor in response to NPY (Figure S13D).

To further compare the pharmacological featured of NPY and [Leu³¹, Pro³⁴]–NPY, we measured the NPY or [Leu³¹, Pro³⁴]–NPY-induced Y₁R and Y₂R activation. For Y₁R, NPY and [Leu³¹, Pro³⁴]–NPY displayed similar potencies on Y₁R or mutants (Figure 5D). However, both NPY analogs exhibited different activation potencies for Y₂R and its mutants (Figure 5E).

In addition, structural comparisons of Y₁R and Y₂R revealed a notable displacement of R³³ from NPY peptides in two receptors (Figure 5F). In Y₂R structure, the residue R³³ is observed to interact with D292^6.59 and forms polar interactions with R³⁵ and E205^ECL2 via hydrogen bonding (Figure 5F). In Y₁R structure, the side chain of R³³ in NPY is found to insert a cavity shaped by TM6 and TM7, establishing hydrophobic contacts with residues F302^7.35, H298^7.31, and F286^6.58 in Y₁R (Figure 5F). Consistent with the structural observations, the results of mutagenesis studies and functional assays showed that the mutation E205^ECL2A influenced NPY induced Y₂R activation potency, whereas the mutation N283^6.55A reduced the Y₁R activation induced by NPY (Figures 5H, S9–12 and Table S2). Notably, the residues F286^6.58 in Y₁R appears to be a key facet to shape the sub-pocket (Figure 5G).

2.6 Structural basis of NPY mediated Y₁R and Y₂R activation

The structural comparison of our two Gi2-coupled NPY receptors with other class A GPCRs revealed similar conformations.⁴⁶ TM6 and TM7 of Y₁R and Y₂R adopted nearly identical conformations to the active structures of the neurotensin receptor NTS1,⁴⁷ orexin receptors,³⁵ and the endothelin ET_B receptor³³ (Figure S14). Furthermore, the structural comparison of two Y₁R and Y₂R receptors with the antagonist-bound NPYRs (PDB ID: 5ZBH, 5ZBQ, and 7DDZ),^{31, 32} supports the contention that these two complexes were in the active state (Figures 6A and D). Compared with structure of antagonist BMS-193885 bound Y₁R (PDB ID: 5ZBH), the activated Y₁R complex displayed pronounced outward displacements (∼8 Å, measured at Cα of T258^6.30) of the TM6 at cytoplasmic region, and ∼5 Å inward shift of TM7 (measured at Cα of Y320^7.53) (Figure 6A).

The residue F286^6.58 in Y₁R was demonstrated to play important role in receptor activation as well as ligand selectivity. The detailed comparison of both active and inactive structures reveals a significant movement of F286^6.58 toward receptor helical core, which may initiate the cascade of conformational changes upon agonist binding. Meanwhile, the microswitch residue W^6.48 is observed to display rotameric change (Figures 6C and F), synergistically, F^6.44 in P-I-F motif underwent conformational changes to facilitate G-protein coupling (Figure S15). In addition, residue Q^3.32 conserved in Y₁R and Y₂R formed hydrogen bonds with NPY, leading to a slight upward movement of TM3 (Figures 6A, B, D, and E).