Molecular basis for ligand recognition and receptor activation of the prostaglandin D2 receptor DP1

Significance

The prostaglandin D2 receptor (DP1) functions as a critical regulator of diverse physiological processes, including sleep–wake cycles, allergic responses, and inflammatory cascades. By determining high-resolution cryo-EM structures of DP1 in multiple functional states, we uncovered an activation mechanism that challenges the classical GPCR paradigm. Our structures reveal ligand-specific activation pathways and unique transmembrane helix interactions that differ markedly from the conventional toggle switch model characteristic of Class A GPCRs. These molecular insights establish a structural foundation for developing next-generation DP1-targeted therapeutics with enhanced selectivity and reduced side effects, addressing a significant unmet need in treating disorders ranging from sleep disturbances to inflammatory diseases.

Abstract

The prostaglandin D2 receptor 1 (DP1), a rhodopsin-like Class A GPCR, orchestrates critical physiological and pathological processes, ranging from sleep regulation to inflammatory responses and cardiovascular function. Despite its therapeutic significance, structural insights into DP1 activation mechanisms have remained elusive. Here, using cryoelectron microscopy (cryo-EM), we determined high-resolution structures of human DP1 in both inactive and active states, with the latter captured in complex with its endogenous agonist PGD2 or the synthetic agonist BW245C, bound to the stimulatory G protein, Gs. Our structures, coupled with functional and mutagenesis studies, unveiled unique structural features of DP1, including an alternative activation mechanism, ligand-selectivity determinants, and G protein coupling characteristics. These molecular insights provide a rational framework for designing selective DP1-targeted therapeutics, both agonists and antagonists, with enhanced specificity and reduced off-target effects, opening broad avenues for treating DP1-associated disorders.

G protein–coupled receptors (GPCRs) constitute the largest family of membrane proteins and are involved in regulating a myriad of physiological processes (1). As such, GPCRs represent major therapeutic targets across many disease areas. The prostaglandin receptor family is a key GPCR subfamily that mediates the biological actions of prostaglandins, a class of lipid mediators derived from membrane phospholipids through the catalytic activity of cyclooxygenases and other enzymes (2, 3). In humans, the major prostaglandins include prostaglandin D2 (PGD₂), prostaglandin E2 (PGE₂), prostaglandin F2α (PGF_2α), prostacyclin (PGI₂), and thromboxane A2 (TXA₂). These bioactive lipids exert their effects by binding to and activating specific prostaglandin receptors, which belong to the rhodopsin-like Class A GPCR family and consist of nine subtypes: prostaglandin D₂ receptors DP1 and DP2, prostaglandin E₂ receptors EP1 to EP4, prostaglandin F_2α receptor FP, prostacyclin receptor IP, and thromboxane receptor TP (4).

Among these receptors, the PGD₂ receptor DP1 has emerged as a key receptor in both the central nervous system (5, 6) and the immune system. Upon stimulated by PGD₂, DP1 primarily couples to the stimulatory G protein G_s to activate the adenylyl cyclase pathway, leading to an increase in intracellular cAMP levels, thereby mediating the regulation of physiological and pathological processes. In the brain, PGD₂ is the most abundant prostaglandin and acts through DP1 to regulate sleep and body temperature homeostasis (7, 8). Notably, PGD₂ cooperates with the neuromodulator adenosine to promote sleep by activating DP1 and inducing the release of sleep-promoting factors (5, 6, 9). Moreover, recent studies have implicated the PGD₂/DP1 pathway in inflammatory responses triggered by sleep deprivation (6). In the immune system, PGD₂ is primarily secreted by mast cells, dendritic cells, and T helper 2 cells (10). By signaling through DP1 and the related DP2 receptor, PGD₂ plays crucial roles in various immune-related conditions, including allergic diseases like asthma (11), pneumonia (12, 13), lupus (14), cancer (15), and cytokine storms induced by viral infections such as SARS-CoV-2 (16, 17).

The development of DP1 agonists and antagonists has garnered significant therapeutic interest due to the receptor’s involvement in multiple pathological conditions. DP1 agonists have shown potential for treating allergic disorders like asthma and rhinitis by modulating immune responses and promoting bronchodilation (11). Conversely, DP1 antagonists have been explored for their potential in treating chronic obstructive pulmonary disease, atopic dermatitis, and allergic rhinitis (18).

Among the nine prostaglandin receptors, active form structures have been reported for most members, including EP2 to EP4 (19), FP (20), (21), and IP (22). However, despite the crucial role of DP1 in regulating diverse physiological processes, its structure has remained elusive, hindering our understanding of ligand recognition, receptor activation, and G protein coupling mechanisms specific to this receptor.

In this study, we report cryo-EM structures of human DP1 in inactive and active states, the latter bound to its endogenous agonist PGD2 and the selective agonist BW245C, in complex with the stimulatory G protein Gs. Combined with mutational and functional studies, our structures provide critical insights into the molecular mechanisms underlying ligand selective recognition, receptor activation, and G protein coupling in the DP1 receptor. These findings not only advance our understanding of DP1 receptor function but also facilitate the rational design of therapeutics targeting this key receptor.

Result

Overall Architecture of DP1.

To facilitate the cryo-EM study of DP1-G_s complexes, we implemented several molecular engineering strategies. First, a BRIL tag was introduced at the N-terminus of DP1, and the C-terminal residues 341–359 were removed (23, 24). We also employed the NanoBiT tethering strategy (25), fusing LgBiT to the DP1 C-terminus and HiBiT to Gβ, to enhance complex stability. Further stabilization was achieved using a dominant-negative Gα_s mutation combined with the nanobody Nb35 (26), which binds at the Gα_s-Gβ interface.

The structures of both the PGD₂-DP1-G_s complex and BW245C-DP1-G_s complex were resolved to high resolutions of 2.72 Å and 2.35 Å, respectively. (Fig. 1 A and B and SI Appendix, Figs. S1 and S2 and Table S1). The high-quality density map facilitated unambiguous model building for the receptor structure encompassing residues 3–335, with the exception of several invisible residues within intracellular loop1 (ICL1) (residues 47–55) and intracellular loop3 (ICL3) (residues 237–255) (Fig. 1 A and B and SI Appendix, Fig. S4). Additionally, the density maps provided clear visualization of PGD₂ and BW245C, and most residues of the G_s heterotrimer (Fig. 1 A and B).

Fig. 1.

Structure determination of DP1 -G_s complexes. (A) Cryo-EM density and cartoon representation of the model of the PGD₂-DP1-Gs complex. DP1 is shown in Indian red, Gα_s in light salmon, Gβ in medium purple, Gγ in forest green, Nb35 in light gray, and PGD₂ in gold. (B) Cryo-EM density and cartoon representation of the model of the BW245C-DP1-G_s complex. DP1 is shown in light sea green, G_s in light salmon, Gβ in medium purple, Gγ in forest green, Nb35 in light gray, and BW245C in orchid. (C) Cryo-EM density map and cartoon representation of the inactive DP1 receptor model fused with BRIL in the third intracellular loop (ICL3). The DP1 receptor is rendered in burly wood coloration, while the BRIL fusion protein is displayed in slate gray. (D) Comparison of the ligand-bound DP1-G_s complex with the EP2-G_s complex (PDB ID: 7CX2), EP3-G_i structure (PDB ID: 8GDC), EP4-G_s structure (PDB ID: 8GDB), and FP-G_q structure (PDB ID: 8IUK) at their receptor parts. DP1 is shown in light sea green, EP2 in violet, EP3 in medium purple, EP4 in pink, and FP in yellow. (E) Detailed diagram comparing the ligand-bound DP1-G_s complex with the EP2 structure (PDB ID: 7CX2), EP3 structure (PDB ID: 8GDC), EP4 structure (PDB ID: 8GDB), and FP-G_q structure (PDB ID: 8IUK) at their receptor parts. DP1 is shown in light sea green, EP2 in violet, EP3 in medium purple, EP4 in pink, and FP in yellow.

To facilitate cryo-EM structural studies of the apo DP1, we employed a strategic molecular engineering approach. Specifically, we replaced the third intracellular loop (ICL3) between transmembrane helices 5 and 6 with the BRIL protein. To enhance structural stability, we further stabilized this construct by introducing an anti-BRIL Fab (FabBRIL) (27) and an anti-Fab nanobody (NbFab) (28) (SI Appendix, Fig. S3). Using these modifications, we successfully determined the apo-DP1 structure by cryo-EM at an overall resolution of 3.41 Å (Fig. 1C).

The overall architecture of the active DP1 receptor features the classic seven-transmembrane fold commonly observed in GPCRs. Compared to the inactive apo-DP1, the agonist-bound DP1-Gs complex is in an active state. Structurally, DP1 shares similarities with other prostaglandin receptors such as EP2-EP4 and FP, exhibiting a Cα RMSD range of 0.93 to 1.48 Å (Fig. 1D). This indicates a close structural relationship within this receptor family.

Despite being activated by the same endogenous ligand, PGD₂, DP1 and DP2 exhibit significant structural differences. The low sequence similarity between DP1 and DP2, with only about 15% identity (29), results in substantial structural divergence. The overall structural comparison reveals a larger Cα RMSD of 4.30 Å, underscoring distinct conformational features between these receptors.

One notable structural characteristic of DP1 is the inward displacement of transmembrane helix 5 (TM5) compared to other prostaglandin receptors (Fig. 1 D and E). Furthermore, Helix 8 (H8) of DP1 is not only the shortest among the prostaglandin receptors but also adopts a distinct orientation (Fig. 1E).

The PGD₂ Binding Pocket in the DP1 Receptor.

The cryo-EM map enabled the unambiguous assignment of PGD₂ within the receptor pocket (Fig. 2 A and B). Similar to other prostaglandins, PGD₂ comprises three distinct parts: a carboxyl group-containing α-chain (Region A), a five-membered ring (Region B), and a hydrophobic ω-chain (Region C) (Fig. 2C).

Fig. 2.

The PGD₂ binding pocket of DP1. (A) Vertical cross-section of the PGD₂ binding pocket in the DP1 receptor. (B) Corresponding interactions that contribute to PGD₂ binding in the DP1 receptor. (C) Region division of PGD₂ and corresponding interactions that contribute to the PGD₂ binding with the DP1 receptor. Hydrogen bonds are depicted as red dashed lines. (D) cAMP activation assay of key mutants in the DP1 receptor that bind to PGD₂. ΔpEC₅₀ = pEC₅₀ of PGD₂ for the specific mutant - pEC₅₀ of PGD₂ for the wild-type (WT) receptor. Data are presented as mean values ± SEM; n = 3 independent samples; significance was determined using One-way ANOVA; n.s. = not significant; *P < 0.05; **P < 0.01; ***P < 0.001.

Region A, containing the carboxyl group, is located in a hydrophilic subpocket near the receptor’s top region. It forms hydrogen bonds with T181^ECL2 and W182 ^ECL2 on the extracellular loop 2 (ECL2), and Y87^2.65 on transmembrane helix 2 (TM2) via the carboxyl group of PGD₂. Additionally, R310^7.40 on TM7 forms salt bridges with the carboxyl group, collectively enhancing ligand recognition (Fig. 2 B and C). Mutation of these residues to alanine in DP1 results in a significant decrease in the potency of PGD₂ to activate the receptor.

Region B is primarily composed of a five-membered ring containing one hydroxyl group and one carbonyl group. This region is situated within lateral pockets formed by TM1, TM2, and TM7, engaging in multiple polar interactions with the receptor. Specifically, the carbonyl group within Region B forms hydrogen bonds with the backbone amine atoms of L26^1.42 and F27^1.43 on TM1, and the side chain of S80^2.58 on TM2 (Fig. 2 B and C). Mutating S80^2.58 to alanine in DP1 caused a dramatic reduction in activity of PGD₂, over 100-fold lower, underscoring the significant contribution of Region B to ligand recognition (Fig. 2D).

Region C, extending beyond the polar pockets of Regions A and B, penetrates into a hydrophobic pocket, fostering hydrophobic interactions with G116^3.36, F115^3.35, and L309^7.39 within the pocket (Fig. 2 B and C). Mutation of these amino acids to alanine in DP1 significantly attenuates the affinity for PGD₂ (Fig. 2D and SI Appendix, Fig. S5). Additionally, the hydroxyl group on the ω-chain forms a hydrogen bond with the highly conserved S316^7.46 in DP1 among prostaglandin receptors. Alanine substitution of S316^7.46 in DP1 attenuates the potency of PGD₂ to activate the receptor (Fig. 2D and SI Appendix, Fig. S5).

The binding residues interacting with PGD₂ in DP1 are highly conserved among other prostaglandin receptors, such as EP2, DP2, and FP (30). This conservation suggests that PGD₂ can cross-activate these receptors, displaying high potency in activating them. The cross-activation is mainly due to the high structural similarity between PGD₂ and other prostaglandins, such as PGE₂ and PGF_2α, and the high sequence similarity among prostaglandin receptors (SI Appendix, Fig. S6), particularly in the ligand-binding pocket (31). As a result, PGD₂ can engage in similar interactions with the conserved residues in these receptors, leading to their activation.

Structural Basis for Selective DP1 Agonism.

BW245C is recognized as a highly selective DP1 agonist (31, 32), although its specific potency on other prostaglandin receptors has not been extensively reported. To better assess the selectivity of PGD2 and BW245C among prostaglandin receptors, we evaluated the potency of these ligands to activate several receptors, including EP1, EP2, EP4, and FP. The results showed that BW245C exhibited lower activity on EP1 and FP compared to PGD₂, while displaying higher activity on EP2 and EP4 (SI Appendix, Fig. S5).

To understand the structural basis for this selectivity, we compared the binding poses of BW245C and PGD₂ in the DP1 receptor. Although structural differences exist, BW245C adopts a binding conformation similar to PGD₂ and interacts with three distinct subpockets within the receptor (Fig. 3 A and D).

Fig. 3.

The BW245C binding pocket of the DP1 receptor and the selective mechanism of BW245C. (A) Vertical cross-section of the BW245C binding pocket in the DP1 receptor. (B) Comparison of the binding patterns of PGD₂ and BW245C. (C) Corresponding interactions that contribute to BW245C binding in the DP1 receptor. (D) Region division of BW245C and corresponding interactions that contribute to the BW245C binding with the DP1 receptor. Hydrogen bonds are depicted as red dashed lines. (E) cAMP activation assay of key mutants in the DP1 receptor that bind to BW245C. ΔpEC₅₀ = pEC₅₀ of BW245C for the specific mutant - pEC₅₀ of BW245C for the WT receptor. Data are presented as mean values ± SEM; n = 3 independent samples; significance was determined using One-way ANOVA; n.s. = not significant; *P < 0.05; **P < 0.01; ***P < 0.001. (F) Distribution of minimal distance of BW245C’s carboxyl group to sidechain oxygen of T181 and Y87. Red and blue lines represent WT and R310A systems, respectively. (G) Sequence alignment of prostanoid receptors. Hydrophobic residues are shown in yellow, polar charged residues in blue, and polar uncharged residues in green.

In Region A, the carboxyl group of BW245C interacts with hydrophilic residues in the upper pocket of the receptor. It forms hydrogen bonds with T181^ECL2 and W182^ECL2 located on ECL2 and Y87^2.65 on TM2. Additionally, the carboxyl group forms a salt bridge with R310^7.40 on TM7. However, the mutation of R310^7.40 to alanine does not significantly alter the ability of BW245C to activate DP1, suggesting compensatory interactions that help maintain its activity (Fig. 3 C–E and SI Appendix, Fig. S5). Molecular dynamics simulations of the R310^7.40A mutant revealed that the carboxyl group of BW245C shifts closer to the oxygen atoms of T181^ECL2 and Y87^2.65, increasing interaction frequency within the <3 Å range (Fig. 3F). This adjustment is facilitated by the flexibility of the saturated α-chain in BW245C, which contrasts with the rigidity of the unsaturated α-chain in PGD₂. The enhanced adaptability allows BW245C to form stabilizing interactions with polar residues such as Y87^2.65 and T181^ECL2. Supporting this observation, the Y87^2.65A-R310^7.40A double mutant leads to a 30-fold reduction in BW245C potency compared to the Y87^2.65A single mutant, highlighting the critical role of Y87^2.65 binding in the absence of R310^7.40 (Fig. 3E and SI Appendix, Fig. S5).

Region B contains the 2,5-dioxoimidazolidin moiety of BW245C, which acts as a bioisosteric substitute for the five-membered ring in PGD₂. This region interacts with residues conserved within lateral subpockets formed by TM1, TM2, and TM7, including L26^1.42, F27^1.43, and S80^2.58 (Fig. 3 C and D). Similar to PGD₂, the alanine mutation of S80^2.58 results in a substantial reduction in BW245C activity, emphasizing the importance of this residue in ligand recognition (Fig. 3E).

Region C includes the cyclohexyl group of BW245C, which enhances hydrophobic interactions with residues such as G116^3.36, F115^3.35, and L309^7.39 in the hydrophobic pocket of DP1. The flexible ω-chain of BW245C, unlike the rigid ω-chain of PGD₂, appears to limit its ability to interact with toggle-switch residues like W^6.48 in EP1 and FP, reducing classical activation (Fig. 3G). However, BW245C likely activates EP2 and EP4 through nonclassical mechanisms, which will be addressed in the following section.

These findings show that PGD₂ and BW245C share overlapping binding features in DP1, but structural differences, particularly in flexibility and interaction dynamics, contribute to their distinct receptor selectivity profiles.

The Active Structure of DP1.

The binding interactions of PGD₂ and BW245C with DP1 facilitate receptor activation. To gain structural insights into this process, we aligned the inactive DP1 structure with the active β₂-adrenergic receptor(26) (β₂AR, PDB: 3SN6).

DP1 activation displays the hallmark class A GPCR characteristics, including outward displacement at the cytoplasmic end of transmembrane helix 6 (TM6), accompanied by inward shifts of TM7 and TM5 (4 Å) relative to inactive DP1 (Fig. 4A). While these conformational changes parallel those observed in β2AR, DP1 exhibits distinct differences: The TM6 displacement is 12.3 Å smaller, whereas the TM5 deflection is more pronounced (Fig. 4A).

Fig. 4.

Active structure of DP1. (A) Comparison of transmembrane helices TM5, TM6, and TM7 of the active DP1 receptor, β₂ adrenergic receptor (β₂AR), and inactive DP1 receptor. (B) Distance display between the ligands and the residue at position 6.48. (C) The steric clashes that exists between TM6 and TM7. (D) Distance display between the ligands and the residue at position 7.47. (E and F) The rotation and the map density of K76 in active and inactive DP1. (G) A steric clash exists between I317 in the inactive DP1 receptor and the backbone atoms of TM1 in the active DP1 conformation. (H) Distance display between the ligands and the residue K76^2.54. (I) Distribution of sidechain heavy atom RMSD for K76^2.54. (J and K) cAMP accumulation assay of WT and K76A, I317A mutants in DP1 with PGD₂ and BW245C. (L) Comparison of the D/ERY motif between the active DP1, inactive DP1 and EP4 receptors.

Unlike most class A GPCRs, DP1 employs an unconventional activation mechanism. The canonical W^6.48 toggle switch is replaced by S278^6.48, which is located over 9 Å from both ligands, indicating no direct interaction. Furthermore, neither ligand directly engages TM6 (Fig. 4B). Structural alignment revealed significant clashes between active-state TM7 and inactive-state TM6, suggesting that TM7 plays a crucial role in driving the outward movement of TM6 (Fig. 4C).

Our mutational analysis revealed the functional importance of key residues in the activation process. Alanine mutations of TM7 residues involved in ligand binding (L309^7.39, R310^7.40, and S316^7.46) reduced ligand potency by 10 to 100-fold. More dramatic effects were observed with K76^2.54A and I317^7.47A mutations, which caused 1,000 to 4,000 folds reductions in potency. Notably, I317^7.47 does not directly interact with either ligand (Fig. 4D), despite its profound impact on signaling. These findings suggest a cooperative mechanism where TM7 conformational changes work in concert with TM1/TM2 structural rearrangements during receptor activation (Fig. 3E and SI Appendix, Fig. S5).

The agonist-bound DP1 structure is more compact than the inactive state, primarily due to polar interactions between TM1 and TM2. Two key residues in this process are K76^2.54 and I317^7.47. Structural alignment demonstrated that K76^2.54 in the active state would clash with the TM7 backbone in the inactive conformation, while I317^7.47 in the inactive state would clash with the TM1 backbone in the active state (Fig. 4 E–G). This indicates that TM1 and TM2 indirectly control TM7 positioning during activation.

K76^2.54 is positioned approximately 4Å from PGD₂ and likely serves dual roles in receptor function (Fig. 4G). Upon activation, K76^2.54 rotates toward TM7 (Fig. 4 E and F). Molecular dynamics simulations revealed increased K76^2.54 side chain flexibility in the absence of PGD₂, with RMSD increasing from 2.19 ± 0.66 Å (ligand-bound) to 2.80 ± 0.36 Å (apo state) (Fig. 4I). This suggests that ligand binding constrains K76^2.54 flexibility and promotes its orientation toward TM7. Functionally, the K76^2.54A mutation abolished PGD₂-induced activation and reduced BW245C potency 1,000-fold, indicating that K76^2.54 may contribute to DP1 activation, alternatively plays a vital role in ligand binding or pocket architecture formation.

The I317^7.47A mutation reduced PGD₂ potency by 100-fold and BW245C potency by more than 4,000-fold (Fig. 4 J and K). This differential effect highlights the critical role of I317^7.47 in maintaining proper TM1-TM7 interactions necessary for receptor activation. The more pronounced effect on BW245C signaling suggests that different ligands may preferentially utilize distinct structural elements within the receptor activation pathway.

Receptor activation in most class A GPCRs involves rearrangement of the conserved E^3.49-R^3.50-Y^3.51 motif, also known as the E/DRY motif, which is critical for activation and signaling. In DP1, this motif is replaced by E129^3.49-C130^3.50-W131^3.51 (Fig. 4L). The absence of the conserved R^3.50, which typically locks the conformation of E^3.49, may contribute to the constitutive activity observed in DP1 (33). Similar mutations in R^3.50 have been shown to cause constitutive activation in other GPCRs, including the β₂AR(34), rhodopsin (35), and the thyrotropin-releasing hormone receptor. In the active DP1 structure, E129^3.49 extends into the G protein binding pocket and directly participates in G protein coupling, potentially explaining the previously reported constitutive activity of DP1(36).

DP1-Gs Coupling.

In the DP1-G_s complex, the Gαs subunit docks into the intracellular pocket of the activated DP1. Comparative structural analyses of the G_s-coupled DP1, β₂AR, and EP2 receptors have revealed differences in the conformations of transmembrane helices 5 and 6 (TM5 and TM6) and the Gα_s subunit within these G_s-coupled GPCR complexes (Fig. 5A). A common feature shared by these receptors, as well as most G_s-coupled GPCRs, is that the length of TM5 is longer than TM6, facilitating efficient G_s coupling (37). Among the three receptors, DP1 exhibits the most significant inward displacement at the cytoplasmic end of TM5. Additionally, the cytoplasmic end of TM6 in DP1 is more inwardly displaced compared to β₂AR and comparable to EP2. To accommodate this inward shift and avoid clashes with the side chains of R231^5.69 and E259^6.29 in DP1, corresponding adjustments in the Gα_s subunit are necessitated, particularly involving the rotation of the N-terminus of the α5 helix of Gα_s away from TM5 (Fig. 5A).

Fig. 5.

DP1-G_s coupling. (A) Comparison of the structures of the Gα_s-coupled DP1 receptor, β₂ adrenergic receptor (β₂AR; PDB ID: 3SN6), and EP2 receptor (PDB ID: 7CX2). The DP1 receptor is shown in light sea green, β₂AR in light purple, and EP2 in pink. (B) The residues in the DP1 receptor, β₂AR, and EP2 receptor that contact the Gα_s subunit. (C) Detailed interactions of the intracellular loop 1 (ICL1) and intracellular loop 2 (ICL2) with the Gα_s subunit. Residues are shown as sticks, with the corresponding cryo-EM density represented as a mesh. (D) Detailed interactions of the “ECW” motif with the Gα_s subunit.

Concomitantly, the Gα_s subunit itself undergoes a 14.1Å movement of the αN helix relative to the β₂AR-G_s complex, positioning it closer to the cytoplasmic terminus of TM4 in the DP1-G_s complex (Fig. 5A). Moreover, the engagement of Gα_s involves exposure to TM1, TM2, TM3, TM5, TM6, TM7, ICL1, ICL2, and Helix 8 of DP1 (Fig. 5B).

Analogous to other G_s-coupled GPCRs, the ICL2 of DP1 orchestrates a diverse array of interactions with the Gα_s subunit, often adopting helical conformations to facilitate varied interactions. A unique feature of prostaglandin receptors is their ability to couple G proteins through TM1 and ICL1(20). In DP1, L41^1.57 in TM1 packs against the L393 in Gα_s subunit, and S59 in ICL1 forms a hydrogen bond with the backbone carbonyl of Q390 in Gα_s (Fig. 5C).

Additionally, in the active structure of DP1, E129^3.49 in the unique E^3.49C^3.50W^3.51 motif extends into the G protein binding pocket and forms hydrogen bonds with Q390 and Y391 in Gα_s (Fig. 5D). This unconstrained E129^3.49 in DP1 plays vital roles in Gα_s coupling and may explain the constitutive activity of DP1 reported previously (33).

Discussion

In this study, we present high-resolution cryo-EM structures of the human DP1 receptor in both inactive and active states, bound to the endogenous agonist PGD₂ and the synthetic agonist BW245C, in complex with the stimulatory G protein Gs. These structures provide critical molecular insights into ligand recognition, receptor activation, and G protein coupling, offering a structural foundation for understanding the unique pharmacological properties of DP1.

Our structural analyses show that PGD₂ and BW245C share overlapping binding sites and cross-activate different prostaglandin receptors. While PGD₂ exhibits high activity on EP1 and FP receptors, BW245C demonstrates higher activity for EP2 and EP4. Both ligands interact with conserved residues in Region A, including T181^ECL2 and W182 ^ECL2, and Y87^2.65 and R310^7.40 within the transmembrane helices. In Region B, the 2,5-dioxoimidazolidin moiety of BW245C mimics the five-membered ring of PGD₂, interacting with residues in lateral subpockets formed by TM1, TM2, and TM7, such as L26^1.42, F27^1.43, and S80^2.58 (Fig. 3 C–E). Notably, the alanine mutation of S80^2.58 significantly reduces BW245C activity, underscoring its critical role in ligand recognition. Unlike PGD₂, which has a rigid ω-chain, the flexible ω-chain of BW245C limits its displacement of the toggle-switch residues like W^6.48 in EP1 and FP, reducing classical activation. However, BW245C likely activates EP2 and EP4 through nonclassical mechanisms. These structural differences help explain the higher potency and selectivity of BW245C over PGD₂ and provide valuable insights for designing improved DP1-selective agonists.

A key finding in our study is the unconventional activation mechanism of DP1, which deviates from the classical “toggle switch” model observed in most Class A GPCRs. While most GPCRs use the conserved W^6.48 residue in their activation, DP1 may employ a cooperative activation mechanism involving both direct ligand-TM7 interactions and TM1/TM2-mediated conformational changes.

Additionally, DP1 lacks the conserved R^3.50 residue in the D/ERY motif, which is replaced by E129^3.49. This substitution contributes to the constitutive activity of DP1 by stabilizing its active conformation and facilitating G protein coupling. E129^3.49 interacts directly with residues in the Gα_s subunit, a feature absent in other prostaglandin receptors. This unique motif, combined with the inward displacement of TM5 and TM6, necessitates structural adjustments in the Gα_s subunit to accommodate the distinctive conformation of DP1. These insights emphasize the structural and functional uniqueness of DP1 among Gs-coupled GPCRs.

Our detailed characterization of the DP1-Gs complex further advances our understanding of G protein coupling. Unlike other Gs-coupled GPCRs, DP1 induces a significant inward displacement of TM5, prompting the α5 helix of Gα_s to rotate and reposition. This adaptation highlights the structural flexibility of GPCR-G protein interactions and provides a framework for understanding the coupling efficiency and signaling specificity of DP1.

In conclusion, our study offers important insights into the molecular basis of DP1 ligand recognition, receptor activation, and G protein coupling. These findings pave the way for the structure-guided design of selective DP1-targeting therapeutics with enhanced efficacy and reduced off-target effects, opening broad avenues for treating DP1-associated diseases such as asthma, allergic rhinitis, and inflammatory disorders.

Materials and Methods

Constructs.

The gene encoding human DP1 (amino acids 1–340) was cloned into a pFastBac vector (Thermo Fisher Scientific) using the ClonExpress II One Step Cloning Kit (Vazyme Biotech). To facilitate expression and stabilization, a prolactin signal peptide and BRIL tag were added to the N-terminus, and an LgBiT tag was fused to the C-terminus. For purification, an OMBP-MBP tag was fused to the C-terminus of LgBiT. A dominant-negative (DN) Gα_s format including mutations S54N, G226A, E268A, N271K, K274D, R280K, T284D and I285T was constructed to reduce the affinity of GDP/GTP -binding and increase the stability of G_s complex. All the three components of G_s-DN, rat Gβ1-SmBit, bovine G_γ2 were also constructed into the pFastBac vector, respectively.

Insect Cell Expression.

The expression of the protein constructs was carried out in High Five insect cells (Invitrogen) using the baculovirus method (Expression Systems). The insect cell cultures were grown in SIM HF Expression Medium (Sino Biological) until reaching a density of 3 to 4 million cells per mL. At this point, the cells were coinfected with five separate baculoviruses, each carrying the respective gene for human DP1, a Gα_s chimera, G_β, and G_γ. After 48 h of infection, the cell culture was collected by centrifugation, and the resulting cell pellets were stored at −80 °C for subsequent purification steps.

Expression and Purification of Nanobody 35.

Nanobody 35 (Nb35) with a C-terminal histidine tag (His6) was expressed in Escherichia coli BL21 (DE3) bacteria. The bacterial culture was grown in Terrific Broth medium supplemented with 100 μg/mL ampicillin until reaching an optical density (OD600) of 0.6 at 37 °C. Protein expression was induced by adding 0.5 mM isopropyl-β-D-thiogalactoside (IPTG), and the culture was further incubated for 16 h at 18 °C. Cells were harvested by centrifugation (4,000 rpm, 20 min), and the Nb35 protein was extracted and purified by nickel-affinity chromatography as previously described (26).

The eluted protein was concentrated and subjected to size-exclusion chromatography using a HiLoad 16/600 Superdex 75 column (GE Healthcare) pre-equilibrated with a buffer containing 20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.4, and 100 mM NaCl. The purified Nb35 was supplemented with 10% (vol/vol) glycerol, flash-frozen in liquid nitrogen, and stored at −80 °C until further use.

Complex Purification.

Cell pellets were thawed in a buffer containing 20 mM HEPES (pH 7.4), 100 mM NaCl, 10 mM MgCl₂, and CaCl₂, 10 μM PGD₂ or BW245C supplemented with a Protease Inhibitor Cocktail (TargetMol). For the purification of the PGD₂/BW245C-DP1-Gα_s-Nb35 complex, membranes were solubilized with 0.5% (w/v) lauryl maltose neopentyl glycol (LMNG) (Anatrace) and 0.1% (w/v) cholesteryl hemisuccinate (CHS) (Anatrace) at 4 °C for 2 h. Insoluble material was removed by centrifugation at 70,000×g for 35 min, and the supernatant was immobilized on an amylose resin (Smart-Lifesciences).

The resin was then packed and washed with 10 column volumes of 20 mM HEPES (pH 7.4), 100 mM NaCl, 0.03% (w/v) LMNG, 0.006% CHS, 10 μM PGD₂ (MCE) or BW245C (MCE). Subsequently, the complex was washed with 10 column volumes of 20 mM HEPES (pH 7.4), 100 mM NaCl, 0.0075% (w/v) LMNG, 0.0025% (w/v) GDN (Anatrace), 0.0015% CHS, 10 μM PGD₂ or BW245C. Finally, the sample was eluted with 10 column volumes of 20 mM HEPES (pH 7.4), 100 mM NaCl, 0.0075% (w/v) LMNG, 0.0025% (w/v) GDN (Anatrace), 0.0015% CHS, 10 mM maltose, and 10 μM PGD₂ or BW245C. The eluent was incubated with Nb35 for 0.5 h.

Complex fractions were concentrated using a 100-kDa molecular weight cut-off (MWCO) Millipore concentrator for further purification. The complex was then subjected to size-exclusion chromatography on a Superose 6 Increase 10/300 GL column (GE Healthcare) pre-equilibrated with a size buffer containing 20 mM HEPES (pH 7.4), 100 mM NaCl, 0.00075% (w/v) LMNG, 0.00025% (w/v) GDN, 0.00015% CHS, and 10 μM PGD₂ or BW245C to separate the complexes. Eluted fractions were evaluated by SDS–PAGE, and those containing the desired complex were pooled and concentrated for cryo-EM experiments.

Inactive DP1 Receptor Complex Purification.

Cell pellets were thawed in a buffer containing 20 mM HEPES (pH 7.4), 150 mM NaCl, 10 mM MgCl2, and CaCl₂, supplemented with a Protease Inhibitor Cocktail (TargetMol). For the purification of the inactive DP1 receptor-FabBRIL-NbFab complex, membranes were solubilized with 0.5% (w/v) lauryl maltose neopentyl glycol (LMNG) (Anatrace) and 0.1% (w/v) cholesteryl hemisuccinate (CHS) (Anatrace) at 4 °C for 2 h. Insoluble material was removed by centrifugation at 70,000×g for 35 min, and the supernatant was immobilized on a nickel affinity resin.

The resin was then packed into a column and washed with 25 column volumes of a buffer containing 20 mM HEPES (pH 7.4), 150 mM NaCl, 0.03% (w/v) LMNG, 0.006% CHS, and 55 mM imidazole. The inactive DP1 receptor was eluted with 10 column volumes of a buffer containing 20 mM HEPES (pH 7.4), 150 mM NaCl, 0.0075% (w/v) LMNG, 0.0025% (w/v) GDN (Anatrace), 0.0015% CHS, and 300 mM imidazole. The eluent was incubated with FabBRIL and NbFab for 2 h.

Complex fractions were concentrated using a 100-kDa molecular weight cut-off (MWCO) Millipore concentrator for further purification. The complex was then subjected to size-exclusion chromatography on a Superose 6 Increase 10/300 GL column (GE Healthcare) pre-equilibrated with a size buffer containing 20 mM HEPES (pH 7.4), 150 mM NaCl, 0.00075% (w/v) LMNG, 0.00025% (w/v) GDN, 0.00015% CHS to separate the complexes. Eluted fractions were evaluated by SDS-PAGE, and those containing the desired complex were pooled and concentrated for cryo-EM experiments.

Cryo-EM Data Collection.

Cryo-EM grids were prepared using the Vitrobot Mark IV plunger (FEI) set at 6 °C and 100% humidity. A 3 mL aliquot of the sample was applied to glow-discharged gold R1.2/1.3 holey carbon grids. The sample was incubated on the grids for 10 s before blotting for 4 s (double-sided, blot force 2) and immediately flash-frozen in liquid ethane.

For the PGD₂-DP1-G_s complex dataset, 7,174 movies were collected on a Titan Krios microscope equipped with a Falcon 4 direct electron detection device operating at 300 kV, with a magnification of 165,000, corresponding to a pixel size of 0.73 Å. Image acquisition was performed using EPU Software (FEI Eindhoven, Netherlands). A total of 36 frames were collected, accumulating to a total dose of 50 e- Å-2 over a 2.5 s exposure.

For the BW245C-DP1-G_s complex dataset, 6,759 movies were collected on a Titan Krios microscope equipped with a G4 direct electron detection device operating at 300 kV, with a magnification of 165,000, corresponding to a pixel size of 0.824 Å. Image acquisition was performed using EPU Software (FEI Eindhoven, Netherlands). A total of 36 frames were collected, accumulating to a total dose of 50 e- Å-2 over a 2.3 s exposure.

Cryo-EM Image Processing.

MotionCor2 was used to perform frame-based motion correction and generate drift-corrected micrographs for further processing (38, 39). All subsequent steps, including contrast transfer function (CTF) estimation, particle picking and extraction, two-dimensional (2D) classification, ab initio reconstruction, heterogeneous refinement, nonuniform refinement, local refinement, and local resolution estimation, were performed using cryoSPARC (40).

For the PGD₂-DP1-G_s dataset, 7,174 dose-weighted micrographs were imported into cryoSPARC, and CTF parameters were estimated using patch-CTF (SI Appendix, Fig. S1). 4,403,231 particles were picked using the template picker from the full set of micrographs and extracted with a pixel size of 1.46 Å. After three rounds of 2D classification, 549,179 particles were selected. Good classes with well-defined features were used to generate a good reference by ab initio reconstruction, while bad classes with poor features were used to generate four bad references by ab initio reconstruction. Using these references, the full set of particles underwent three rounds of heterogeneous refinement, and 148,598 good particles were extracted with a pixel size of 0.73 Å. A 2.83 Å map was reconstructed from these particles by nonuniform refinement. The 549,179 particles selected from 2D classification were extracted with a pixel size of 0.73 Å and subjected to four rounds of heterogeneous refinement with the replaced good reference. Finally, 140,051 particles were selected and used for nonuniform refinement and local refinement, generating a 2.72 Å map.

For the BW245C-DP1-G_s dataset, 6,759 dose-weighted micrographs were imported into cryoSPARC, and CTF parameters were estimated using patch-CTF. A total of 6,563 micrographs with CTF resolution better than 3.5 Å were selected (SI Appendix, Fig. S2). Initial particle selection was performed using the Blob picker on a few micrographs, followed by 2D classification to generate good templates. Subsequently, 5,088,272 particles were picked using the template picker from the full set of micrographs and extracted with a pixel size of 1.648 Å. After three rounds of 2D classification, 693,791 good particles were selected for further heterogeneous refinement. The final map from the PGD₂-DP1-G_s dataset was imported as a good reference. Using these references, the full set of particles underwent three rounds of heterogeneous refinement, and 116,634 selected particles were re-extracted with a pixel size of 0.824 Å. A 2.83 Å map was reconstructed from these particles by nonuniform refinement. The 693,791 particles selected from 2D classification were extracted with a pixel size of 0.824 Å and subjected to four rounds of heterogeneous refinement with the replaced good reference. Finally, 224,658 particles were selected and used for nonuniform refinement and local refinement, generating a 2.35 Å map.

For the apo inactive DP1 dataset, a total of 8,576 dose-weighted micrographs were imported into cryoSPARC, where CTF parameters were estimated using the patch-CTF method. From this initial set, 8,563 micrographs with a CTF resolution better than 3.5 Å were selected (SI Appendix, Fig. S3). Initial particle selection was performed using the Blob picker on a subset of micrographs, followed by 2D classification to generate effective templates. Subsequently, a total of 4,568,341 particles were picked using the template picker from the entire set of micrographs and extracted with a pixel size of 1.46 Å. After five rounds of 2D classification, 367,742 high-quality particles were selected for ab initio reconstruction, which generated a reliable 3D reference. A small number of poor-quality particles were also included to create random maps as negative references. Following three rounds of heterorefinement, 68,971 particles were extracted with a pixel size of 0.73 Å for further nonuniform refinement, resulting in a 4.41 Å map. These particles were then used as templates for Topaz particle picking. The extracted particles underwent one round of 2D classification, and the selected particles were subjected to four additional rounds of heterorefinement using updated reference maps. This process resulted in the selection of 41,270 particles, which were then used for nonuniform refinement and local refinement, yielding a 3.49 Å map. A focused local refinement was conducted using a mask centered on DP1 and fused BRIL, resulting in a final map resolution of 3.41 Å.

Model Building.

A predicted DP1 structure from AlphaFold2 was used as the starting reference model for receptor building (41). Structures of G_s, Gβ, Gγ, and the Nb35 were derived from the PDB entry 7XZ5 (42) and rigid-body fitted into the density. All models were initially fitted into the EM density map using UCSF Chimera (43), followed by iterative rounds of manual adjustment and automated rebuilding in COOT (44) and PHENIX (45), respectively. The model was finalized by rebuilding in ISOLDE (46), followed by refinement in PHENIX with torsion-angle restraints to the input model. The final model statistics were validated using the Comprehensive validation (cryo-EM) tool in PHENIX (45) and provided in SI Appendix, Table S1. All structural figures were prepared using Chimera (43), Chimera X (47) and PyMOL (Schrödinger, LLC).

Glo-Sensor cAMP Assay.

The full-length DP1/EP2/EP4 receptor was cloned into the pcDNA6.0 vector (Invitrogen) with an N-terminal FLAG tag. One day before transfection, AD293 cells were seeded in a 12-well plate with DMEM containing 1% dialyzed fetal bovine serum (FBS). After overnight growth, the cells were transfected with a plasmid mixture consisting of the DP1/EP2/EP4 construct and the cAMP biosensor GloSensor22F (Promega) at a 1:1 ratio. Following 24 h of transfection, the cells were adapted to CO2-independent media containing 2% GloSensor cAMP Reagent (Promega) and plated onto a 384-well assay plate at a density of 4,000 cells/10 μL/well. After incubating for 1 h, an additional 5 μL of buffer containing various concentrations of test compounds was added to each well.

The cells were then measured for luminescence using an EnVision plate reader without further incubation. A nonlinear regression analysis was performed using the sigmoidal dose–response function in GraphPad Prism to calculate the values of maximum response (Emax) and half-maximum effective concentration (EC50) for the test compounds.

Inositol Phosphate Accumulation Assay.

The IP-one HTRF assay kit was used to directly measure the production of IP1 in cultured cells. The full-length EP1/FP receptor was cloned into the pcDNA6.0 vector (Invitrogen) with an N-terminal FLAG tag. One day before transfection, AD293 cells were seeded in a 12-well plate with DMEM containing 1% dialyzed FBS. Seed AD293 cells were transfected with plasmids encoding receptors. After 24 h incubation, the cells were collected and suspended with a stimulation buffer. The cell suspension was then dispensed into a white 384-well plate at a volume of 7 μL (4900 cells) per well. Subsequently, 7 μL of ligands were added into the corresponding wells. After incubation at 37 °C for 1 h, IP-one-d2 and anti-IP-one cryptate Tb conjugate were added sequentially at 3 μL per well and incubated for 30 min at room temperature. IP-one measurement was carried out with EnVision multiplate reader (PerkinElmer) according to the manufacturer’s instructions. Data were normalized to the baseline response of the ligand.

Surface Expression Analysis.

The cell-surface expression levels of WT and mutant DP1 receptors were monitored using a fluorescence-activated cell sorting assay. Briefly, AD293 cells expressing Flag-tagged DP1 constructs were harvested 24 h after transfection. The cells were incubated with a mouse anti-Flag-FITC antibody (Sigma) at a dilution of 1:200 for 2 h at 4 °C in the dark. After incubation, 100 μL of PBS was added to the cells. Finally, the surface expression of DP1 was monitored by detecting the fluorescent intensity of FITC using a Guava® easyCyte flow cytometer. The data were normalized to the WT DP1 expression level. The experiments were performed at least three times, and the data are presented as means ± SEM.

MD Simulations.

The simulation systems were built from the DP1-PGD₂ and DP1-BW245C complexes, with G proteins excluded and missing loops modeled using the Molecular Operating Environment. The R301A mutation was introduced via PyMOL. For the apo system, PGD₂ was removed. Each complex was embedded into a 75 × 75 Å POPC lipid bilayer using packmol-memgen and surrounded by a 12 Å water layer (48). Ionic strength was adjusted to 0.15 mol/L NaCl, with counterions added to neutralize the system.

The FF19SB, Lipid21, and GAFF2 force fields were employed for amino acids, lipids, and ligands, respectively (49–51). Systems underwent energy minimization, followed by heating and equilibration based on established protocols (52, 53). Three independent production runs of 500 ns each were performed using pmemd.cuda in Amber22 under the NVT ensemble at 300 K and 1 atm. Long-range electrostatic interactions were handled via the Particle Mesh Ewald method, while a 10 Å cutoff was applied for short-range electrostatic and van der Waals interactions. The SHAKE algorithm and hydrogen mass repartitioning constrained hydrogen bonds, enabling a timestep of 4 fs (54). RMSD, minimal distances, and representative structures were analyzed using CPPTRAJ (55).

Statistics.

Data from functional studies were analyzed using GraphPad Prism 9.0 (Graphpad Software Inc.). Results are presented as mean ± SEM of three independent experiments with triplicate measurements. Significance was determined with one-way ANOVA and *P < 0.05, **P < 0.01, and ***P < 0.001 vs. WT was considered statistically significant, n.s. vs. WT was considered statistically not significant.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Cryo-EM maps have been deposited in the Electron Microscopy Data Bank under accession codes: EMD-60514 for PGD2-DP1-Gs complex (56); EMD-60513 for BW245C-DP1-Gs complex (57); and EMD-64550 for inactive apo-DP1 (58). The atomic coordinates have been deposited in the Protein Data Bank under accession codes: 8ZW0 for PGD2-DP1-Gs complex (59); 8ZVZ for BW245C-DP1-Gs complex (60); and 9UWD for inactive apo-DP1 (61). All other data are included in the article and/or SI Appendix.