Blockade of signaling through the angiotensin II type 1 receptor (AT1R), a prototypical G protein-coupled receptor (GPCR), by angiotensin receptor blockers (ARBs) is a major therapeutic approach to treating a wide variety of cardiovascular and renal diseases 1 . Like most GPCRs, the AT1R signals through two transducers, G proteins and β-arrestins 2,3 . Previous reports have described β-arrestin-biased peptide orthosteric agonists for the AT1R with potential therapeutic advantages over currently available unbiased ARBs 4-6 . Here we report the DNA- encoded library screening-guided isolation and pharmacological characterization of the first small molecule AT1R allosteric ligands. We use cryo-electron microscopy, double electron- electron resonance spectroscopy, molecular dynamics simulations, and targeted mutagenesis to determine their binding sites, binding modes and conformational mechanisms driving their unique and divergent modulatory effects on G protein and β-arrestin pathways. Our findings uncover new mechanisms for precisely controlling the dynamic behavior of the AT1R with implications for drug development targeting this pathophysiologically important receptor family.

Bitter taste functions as a means of both protection against potentially toxic compounds and savoring bitter tasting foods and beverages. Among the 26 bitter taste receptors, taste receptor type 2 member 43 (TAS2R43) has been identified as key for recognizing the bitter taste of coffee. TAS2R43 has also been implicated in many other physiological processes, including the regulation of glucagon-like peptide 1 release from the intestine, bronchodilation, innate immunity and metabolism. Here we report cryo-electron microscopy structures of human TAS2R43 coupled with inhibitory G protein or gustducin (Ggust) stabilized by the potent nephrotoxin and carcinogen aristolochic acid I. Both structures revealed that aristolochic acid I binds in a presumed orthosteric pocket shared with other bitter taste receptor. Further structural, functional and computational studies revealed potential modes for coffee's constituents including caffeine and cafestol, which are bitter tastants from coffee. Lastly, long-timescale molecular dynamics simulations identified potential cryptic allosteric pockets in TAS2R43. These structures could accelerate the search for specific bitter taste ligands that ultimately may be therapeutically useful.

Opioid receptors signal through Gi/o protein and β-arrestin pathways that mediate distinct effects of opiate drugs. While opioid binding and G protein activation are well studied, β-arrestin recruitment remains poorly understood. Here, we determine the complex structure of the kappa opioid receptor (KOR) with β-arrestin1 (βarr1) at 2.60 Å resolution using cryogenic electron microscopy. Structural and mass spectrometry analyses reveal multiple phosphorylation sites and a phospholipid-binding site that specifically enhances arrestin recruitment. The KOR-βarr1 complex adopts a core interaction and exhibits notable differences from other GPCR-βarr1 complexes. Comparisons with the structures of KOR-Nb39 and KOR-Gi1 complexes also reveal distinct structural features in the orthosteric binding site and the KOR-transducer interface that affect signaling bias. Using extensive 3D variation analysis and molecular dynamics simulations, we identify a range of conformational dynamics in both the receptor and βarr1, suggesting an allosteric pathway for arrestin's entry and exit.

Ketamine offers rapid relief for treatment-resistant depression and severe pain in the clinic, providing immediate benefits that traditional medications often fail to deliver. While its antagonistic action at the N-methyl-D-aspartate receptor (NMDAR) is a key mechanism, ketamine's dual nature as both a promising treatment and a drug with abuse potential suggests its therapeutic effects extend beyond NMDAR inhibition. Here we provide structural evidence of human opioid receptors bound to ketamine and its parent analog phencyclidine (PCP), supporting that both ligands can directly bind and activate opioid receptors. The structures, together with site-directed mutagenesis and structure-activity relationship studies, identify key motifs involved in ketamine and PCP recognition and efficacy modulation. Furthermore, we determine the structure of the ligand-free state of human κ opioid receptor, revealing molecular details before ligand engagement. Compared to PCP, ketamine displays more notable binding dynamics in the orthosteric site that may contribute to its unique pharmacology at opioid receptors. Our findings highlight the importance of including opioid receptors to fully understand ketamine's versatility in clinical settings.

The ubiquitous CLC membrane transporters are unique in their ability to exchange anions for cations. Despite extensive study, there is no mechanistic model that fully explains their 2:1 Cl‒/H+ stoichiometric exchange mechanism. Here, we provide such a model. Using differential hydrogen-deuterium exchange mass spectrometry, cryo-EM structure determination, and molecular dynamics simulations, we uncovered conformational dynamics in CLC-ec1, a bacterial CLC homolog that has served as a paradigm for this family of transporters. Simulations based on a cryo-EM structure at pH 3 revealed critical steps in the transport mechanism, including release of Cl‒ ions to the extracellular side, opening of the inner gate, and water wires that facilitate H+ transport. Surprisingly, these water wires occurred independently of Cl‒ binding, prompting us to reassess the relationship between Cl‒ binding and Cl‒/H+ coupling. Using isothermal titration calorimetry and quantitative flux assays on mutants with reduced Cl‒ binding affinity, we conclude that, while Cl‒ binding is necessary for coupling, even weak binding can support Cl‒/H+ coupling. By integrating our findings with existing literature, we establish a complete and efficient CLC 2:1 Cl‒/H+ exchange mechanism.

A common starting point for drug design is to find small chemical groups or "fragments" that form interactions with distinct subregions in a protein binding pocket. The subsequent challenge is to assemble these fragments into a molecule that has high affinity to the protein, by adding chemical bonds between atoms in different fragments. This "molecule assembly" task is particularly challenging because, initially, fragment positions are known only approximately. Prior methods for spatial graph completion-adding missing edges to a graph whose nodes have associated spatial coordinates-either treat node positions as fixed or adjust node positions before predicting edges. The fact that these methods treat geometry and topology prediction separately limits their ability to reconcile noisy geometries and plausible connectivities. To address this limitation, we introduce EdGr, a spatial graph diffusion model that reasons jointly over geometry and topology of molecules to simultaneously predict fragment positions and inter-fragment bonds. Importantly, predicted edge likelihoods directly influence node position updates during the diffusion denoising process, allowing connectivity cues to guide spatial movements, and vice versa. EdGr substantially outperforms previous methods on the molecule assembly task and maintains robust performance as noise levels increase. Beyond drug discovery, our approach of explicitly coupling geometry and topology prediction is broadly applicable to spatial graph completion problems, such as neural circuit reconstruction, 3D scene understanding, and sensor network design.

The free fatty acid receptor 2 (FFA2) is a G protein-coupled receptor (GPCR) that selectively recognizes short-chain fatty acids to regulate metabolic and immune functions. As a promising therapeutic target, FFA2 has been the focus of intensive development of synthetic ligands. However, the mechanisms by which endogenous and synthetic ligands modulate FFA2 activity remain unclear. Here, we present the structures of the human FFA2-Gi complex activated by the synthetic orthosteric agonist TUG-1375 and the positive allosteric modulator/allosteric agonist 4-CMTB, along with the structure of the inactive FFA2 bound to the antagonist GLPG0974. Structural comparisons with FFA1 and mutational studies reveal how FFA2 selects specific fatty acid chain lengths. Moreover, our structures reveal that GLPG0974 functions as an allosteric antagonist by binding adjacent to the orthosteric pocket to block agonist binding, whereas 4-CMTB binds the outer surface of transmembrane helices 6 and 7 to directly activate the receptor. Supported by computational and functional studies, these insights illuminate diverse mechanisms of ligand action, paving the way for precise GPCR-targeted drug design.

As a key mitochondrial Ca2+ transporter, NCLX regulates intracellular Ca2+ signalling and vital mitochondrial processes1-3. The importance of NCLX in cardiac and nervous-system physiology is reflected by acute heart failure and neurodegenerative disorders caused by its malfunction4-9. Despite substantial advances in the field, the transport mechanisms of NCLX remain unclear. Here we report the cryo-electron microscopy structures of NCLX, revealing its architecture, assembly, major conformational states and a previously undescribed mechanism for alternating access. Functional analyses further reveal an unexpected transport function of NCLX as a H+/Ca2+ exchanger, rather than as a Na+/Ca2+ exchanger as widely believed1. These findings provide critical insights into mitochondrial Ca2+ homeostasis and signalling, offering clues for developing therapies to treat diseases related to abnormal mitochondrial Ca2+.

The function of biomolecules such as proteins depends on their ability to interconvert between a wide range of structures or conformations. Researchers have endeavored for decades to develop computational methods to predict the distribution of conformations, which is far harder to determine experimentally than a static folded structure. We present ConforMix, an inference-time algorithm that enhances sampling of conformational distributions using a combination of classifier guidance, filtering, and free energy estimation. Our approach upgrades diffusion models---whether trained for static structure prediction or conformational generation---to enable more efficient discovery of conformational variability without requiring prior knowledge of major degrees of freedom. ConforMix is orthogonal to improvements in model pretraining and would benefit even a hypothetical model that perfectly reproduced the Boltzmann distribution. Remarkably, when applied to a diffusion model trained for static structure prediction, ConforMix captures structural changes including domain motion, cryptic pocket flexibility, and transporter cycling, while avoiding unphysical states. Case studies of biologically critical proteins demonstrate the scalability, accuracy, and utility of this method.

The mitochondrial pyruvate carrier (MPC) governs the entry of pyruvate-a central metabolite that bridges cytosolic glycolysis with mitochondrial oxidative phosphorylation-into the mitochondrial matrix1-5. It thus serves as a pivotal metabolic gatekeeper and has fundamental roles in cellular metabolism. Moreover, MPC is a key target for drugs aimed at managing diabetes, non-alcoholic steatohepatitis and neurodegenerative diseases4-6. However, despite MPC's critical roles in both physiology and medicine, the molecular mechanisms underlying its transport function and how it is inhibited by drugs have remained largely unclear. Here our structural findings on human MPC define the architecture of this vital transporter, delineate its substrate-binding site and translocation pathway, and reveal its major conformational states. Furthermore, we explain the binding and inhibition mechanisms of MPC inhibitors. Our findings provide the molecular basis for understanding MPC's function and pave the way for the development of more-effective therapeutic reagents that target MPC.

While generative AI is transforming the de novo design of proteins, its effectiveness for structure-based design of small molecules remains limited. Current methods, including diffusion models, often produce small molecules with difficult-to-synthesize structures, poor medicinal chemistry properties, and limited target selectivity. To address these limitations, we introduce MedSAGE, a novel generative AI framework that adapts diffusion models specifically for de novo small-molecule design. Rather than using atoms or strings, we develop a novel representation for generating molecules using fragments relevant for medicinal chemistry. Chemical and geometric information characterizing these fragments is embedded in a smooth and interpretable latent space. We also develop an algorithm to optimize the connectivity between generated fragments while preserving chemical validity and synthesizability. In a benchmark of multiple methods across 25 therapeutically relevant protein targets, MedSAGE achieved state-of-the-art performance, producing synthesizable, drug-like molecules with predicted affinity and selectivity closely matching known drugs and drug candidates. Compared to large-scale virtual screening, MedSAGE produced molecules with high predicted affinity over 100 times more efficiently. Our results demonstrate that MedSAGE is already practically useful and paves the way for next-generation tools in structure-guided drug design.

Atomic-level simulations are widely used to study biomolecules and their dynamics. A common goal in such studies is to compare simulations of a molecular system under several conditions—for example, with various mutations or bound ligands—in order to identify differences between the molecular conformations adopted under these conditions. However, the large amount of data produced by simulations of ever larger and more complex systems often renders it difficult to identify the structural features that are relevant to a particular biochemical phenomenon. We present a flexible software package named Python ENSemble Analysis (PENSA) that enables a comprehensive and thorough investigation into biomolecular conformational ensembles. It provides featurization and feature transformations that allow for a complete representation of biomolecules such as proteins and nucleic acids, including water and ion binding sites, thus avoiding the bias that would come with manual feature selection. PENSA implements methods to systematically compare the distributions of molecular features across ensembles to find the significant differences between them and identify regions of interest. It also includes a novel approach to quantify the state-specific information between two regions of a biomolecule, which allows, for example, tracing information flow to identify allosteric pathways. PENSA also comes with convenient tools for loading data and visualizing results, making them quick to process and easy to interpret. PENSA is an open-source Python library maintained at https://github.com/drorlab/pensa along with an example workflow and a tutorial. We demonstrate its usefulness in real-world examples by showing how it helps us determine molecular mechanisms efficiently.

Recent advances in protein structure prediction have highlighted the importance of a longstanding problem: given multiple structural models of a protein, how does one select the best model to use when predicting interactions between that protein and candidate drug molecules? Here we demonstrate the value of a previously unutilized source of information in addressing this problem. We show that given multiple ligands known to bind the protein, one can perform effective model selection by comparing the predicted binding poses of multiple ligands at each model. We introduce a method, RevBind, that exploits this information, leveraging the statistical tendency of different ligands to form similar chemical interactions with a protein's binding pocket. RevBind can be used, for example, to select among variants of AlphaFold models, identifying those that are most useful for molecular docking. Our findings pave the way for the development of even better model selection methods that draw simultaneously on the information used by RevBind and the information used by previous methods.