ChEMBL database molecular pairs
From the PostgreSQL version of CHEMBL34 (ref. 4) (https://doi.org/10.6019/CHEMBL.database.34), we filtered the database for compounds with activities reported against a single protein, with either Ki, Kd, IC50 or EC50 as the activity type. To be considered a molecular pair, compounds had to have the same assay ID, the same target ID, the same activity type and the same reference document (publication) ID, in addition to having a parent–analogue relationship as defined in this work, meaning a C–H group replaced by C–OH, C–F, C–Cl, C–Br, C–CH3 or an aromatic carbon replaced by a nitrogen. This left us with 191,732 parent–analogue pairs.
Analogue enumeration and synthesis
Eighteen parent compounds were selected from previously published literature51,59,60 or datasets (https://asapdiscovery.org/outputs/molecules/#ASAP-SARS-COV-2-NSP3-MAC1) based on their known binding activity, structural relevance or representation of diverse chemical scaffolds. Starting from a parent compound, we used RDKit (http://www.rdkit.org/) to identify all C–H bonds and iteratively replaced the hydrogen atom with a methyl, hydroxyl, fluoro, chloro or bromo group. We also identified aromatic carbons with two heavy-atom neighbours and replaced them with nitrogen. Every analogue generated was represented as a canonical isomeric SMILES string and added to a set to remove duplicates. For each of the 18 parents, the full set of possible analogues that could be synthesized for less than US$400 for 10 mg was ordered.
Confidence intervals and statistics
When reported, 95% confidence intervals were derived from bootstrap resampling with 10,000 iterations. In cases in which the observed frequency of success is exactly 0, bootstrap will fail to give an upper bound for the interval. In such cases, we used the Clopper–Pearson method to estimate an upper bound based on the sample size61. Pearson and Spearman correlations with their associated P values were computed using the scipy.stats module from the SciPy package62.
FEP simulations
These were conducted using FEP+ within the Schrödinger software suite (v2025-2) with the OPLS4 force field63 and the modified simple point charge water model. The default setting was used for the number of lambda windows selection where it depends on the type of perturbations; charge-changing, core hopping and all other perturbations have 24, 16, and 12 lambda windows, respectively. For alchemical transformations with charge changes, the total charge of the simulation box was kept constant by transmuting a Na+ or Cl− ion to water or vice versa. In addition, a 0.15 M concentration of NaCl was added to the simulation box of charge-changing perturbations. For α2A, CB2 and SERT, the FEP+ membrane protocol was applied where a POPC membrane was added to the system in simulation. All other settings were kept default except that the simulation time was extended from 5 ns to 10 ns.
The default FEP map generation protocol was used with the parent compound selected as the biased node. In preparing proteins and ligands for FEP+, the Schrödinger protein preparation workflow and LigPrep were used. The initial binding poses of parent compounds were from poses generated by DOCK3.8, and analogues were aligned to the parents with severe steric clashes resolved using the FEP+ Pose Builder workflow.
Docking
Whereas no structural information was used in the design of analogues from parent compounds, for FEP+ calculations and for post hoc structural analysis, we generated ligand-bound complexes of the parents and relevant ligands using DOCK.3.8 (refs. 54,64). Ligands were docked into the receptor-binding site using grids prepared in previous studies51,59,60,65,66.
Assay selection
We used one consistent readout per parent series (no mixing within a series). SERT and AmpC are reported as Ki; Mac1 is reported as IC50 from peptide displacement (a scalable functional hydrolysis assay is unavailable); GPCR agonist series are reported as EC50 and antagonist series as Ki, such that μOR is EC50 only, whereas α2A and CB2 include EC50 and Ki depending on the parent series. Representative concentration–response curves and the corresponding Z values for each assay are provided in Supplementary Fig. 3.
Transporter assays for SERT K
i
SERT activity was measured using the Neurotransmitter Transporter Uptake Assay Kit from Molecular Devices (#R8174), following the manufacturer’s protocol with slight modifications as previously described65. HEK293T cells (ATCC CRL-3216) stably expressing human SERT were plated in poly-L-lysine-coated 384-well black, clear-bottom plates at a density of 15,000 cells in 40 µl per well, using DMEM supplemented with 1% dialysed FBS. Cells were incubated overnight at 37 °C with 5% CO2 to allow adherence and recovery. The following day, the medium was removed by flicking, and cells were incubated with 25 µl per well of test compound solutions prepared in assay buffer (1× HBSS, 20 mM HEPES, pH 7.4, supplemented with 1 mg ml−1 BSA) for 30 min at 37 °C. After drug treatment, 25 µl per well of dye solution (as provided in the kit) was added directly to the wells, followed by an additional 30-min incubation at 37 °C. Fluoxetine (10 µM) was used as a positive control for SERT inhibition. Fluorescence was measured using the FlexStation II microplate reader with excitation at 440 nm and emission at 520 nm. Relative fluorescence units were exported and analysed using GraphPad Prism 10.0 to calculate IC50 values, from which Ki values were derived using the Cheng–Prusoff equation.
CB2 radioligand-binding assay
CB2 receptor-binding assays were performed using membrane preparations from HEK293 cells (ATCC CRL-1573) stably expressing human CB2, following previously published methods67,68. Membranes were resuspended in TME buffer containing 0.1% BSA (w/v) and 25 µg of membrane protein was added per well. The assay was conducted using the radioligand [3H]CP-55,940 at a final concentration of 0.75 nM, prepared in assay buffer. Nonspecific binding was defined in the presence of 5 µM unlabelled CP-55,940. Test compounds were applied at increasing concentrations to assess competition. Reactions were incubated at 30 °C for 1 h with gentle shaking. After incubation, samples were transferred to Unifilter GF/B 96-well filter plates and filtered using a Packard Filtermate-196 cell harvester (PerkinElmer). Plates were washed four times with ice-cold wash buffer (50 mM Tris-HCl, 5 mM MgCl2 and 0.5% BSA, pH 7.4). Radioactivity bound to the filters was quantified via liquid scintillation counting. Specific binding was calculated by subtracting nonspecific binding from total binding. IC50 and Ki values were calculated using nonlinear regression in GraphPad Prism 9 using the Cheng–Prusoff equation.
α2A receptor-binding assay
α2AAR binding was performed using membrane preparations from insect cells (Expression Systems, 94-001S) expressing human α2A receptors, as previously described51. Membranes were incubated with increasing concentrations of test compounds and 5 nM [3H]rauwolscine in buffer containing 20 mM HEPES (pH 7.5) and 100 mM NaCl, at room temperature for 2 h. After incubation, samples were filtered onto GF/B filter plates, washed with ice-cold buffer and radioactivity was quantified by liquid scintillation counting. IC50 and Ki values were derived using nonlinear regression in GraphPad Prism.
GloSensor cAMP assay for α2A, CB2 and MOR
The GloSensor cAMP assay was performed following the manufacturer’s instructions (Promega) with slight modifications as previously described51,69. In brief, wild-type human α2A, CB2 and μOR were cloned into the pcDNA3.1 vector and co-transfected with the 22F cAMP GloSensor plasmid into HEK293T cells (ATCC CRL-3216) cultured in six-well plates. After 24 h, cells were reseeded into 96-well white plates in CO2-independent medium and equilibrated with GloSensor cAMP reagent as per the manufacturer’s protocol. Cells were incubated for 1 h at 37 °C followed by 1 h at room temperature. Where applicable, 10 μM forskolin was used to elevate basal cAMP levels for assessing receptor-mediated inhibition. Serially diluted test compounds were added, and luminescence signals were recorded using a PerkinElmer microplate reader. Data were analysed using GraphPad Prism 9.0 to calculate EC50 or IC50 values.
AmpC β-lactamase inhibition assay
The AmpC β-lactamase inhibition assay was performed as previously described60. The candidate inhibitors were dissolved in DMSO (20 mM stock) and diluted to maintain a constant 1% DMSO (v/v) in 50 mM sodium cacodylate buffer (pH 6.5). Assays were performed in the presence of 0.01% Triton X-100 to reduce aggregation artefacts. AmpC enzymatic activity was monitored spectrophotometrically using CENTA or nitrocefin as substrates. Initial screening was performed at 200 µM, 100 µM and 40 µM compound concentrations. Substrate concentrations were selected based on known Km values to achieve defined [S]:Km ratios: for CENTA ([S] = 50 µM, Km = 27.6 µM) and nitrocefin ([S] = 100 µM or 28 µM, Km = 180 µM). Reactions were carried out in 96-well format on a BMG Labtech CLARIOstar plate reader, with substrate and enzyme injected into wells containing the inhibitor, followed by kinetic measurement over 50 s. IC50 values were determined by fitting inhibition curves in GraphPad Prism using a fixed Hill coefficient of 1, and Ki values were calculated using the Cheng–Prusoff equation70.
HTRF assay for Mac1
Binding of the compounds to Mac1 was assessed by the displacement of an ADPr-conjugated biotin peptide from His6-tagged protein using a homogeneous time-resolved fluorescence (HTRF)-based assay, as previously described71. The expression sequences used for SARS-CoV-2 Mac1 are listed below. All proteins were expressed and purified as previously described for SARS-CoV-2 Mac1 (ref. 71). Compounds were dispensed into ProxiPlate-384 Plus (PerkinElmer) assay plates using an Echo 650 Liquid Handler (Beckman Coulter). Binding assays were conducted in a final volume of 16 μl with 12.5 nM NSP3 Mac1 protein, 200 nM peptide ARTK(Bio)QTARK(Aoa- RADP)S (Cambridge Peptides), 1:20,000 anti-His6-Eu3+ cryptate (HTRF donor; PerkinElmer AD0402) and 1:500 streptavidin-XL665 (HTRF acceptor; PerkinElmer 610SAXLB) in assay buffer (25 mM HEPES pH 7.0, 20 mM NaCl, 0.05% bovine serum albumin and 0.05% Tween-20, the latter also to reduce aggregation artefacts). Assay reagents were dispensed manually into plates using an electronic multichannel pipette. Mac1 and peptide were pre-incubated for 30 min at room temperature before HTRF reagents were added. Fluorescence was measured after a 1-h incubation at room temperature using a Perkin Elmer EnVision 2105-0010 Dual Detector Multimode microplate reader with dual emission protocol (A = excitation of 320 nm and emission of 665 nm, and B = excitation of 320 nm and emission of 620 nm). Compounds were tested in triplicate in a 14-point dose response. Raw data were processed to give an HTRF ratio (channel A/B × 10,000), which was used to generate IC50 curves using nonlinear regression using GraphPad Prism v10.0.2 (GraphPad Software).
hERG channel inhibition
hERG channel inhibition was evaluated using a Thallium Flux assay on HEK293 cells (ATCC CRL-1573) stably expressing the hERG potassium channel. Cells were seeded at a density of 8,000 cells per well in 384-well poly-D-lysine-coated plates and incubated for 24 h under standard conditions (37 °C at 5% CO2). The next day, a thallium-sensitive dye was added to the cells, followed by a 1-h incubation to ensure dye uptake. Test compounds were added to achieve a final concentration of 30 μM in 0.5% DMSO, and the cells were incubated for an additional 30 min at room temperature. Subsequently, a stimulation buffer containing thallium was added, and fluorescence measurements were taken using a FLIPR Tetra system. Data were collected every 3 s for 3 min (excitation of 470–495 nm and emission of 515–575 nm). Fluorescence intensity over time was analysed to calculate the area under the curve from which percentage inhibition was determined versus haloperidol at 100 μM (positive control) and vehicle (DMSO).
α2A receptor purification and structure determination
Wild-type human α2AAR was cloned into a pVL1392 vector with an N-terminal FLAG tag. The construct was expressed in Spodoptera frugiperda (Sf9) insect cells using the BestBac system. Cells at a density of 4 × 106 cells per millilitre were infected with virus and incubated for 48 h at 27 °C. The receptor was solubilized and purified by FLAG affinity chromatography and size-exclusion chromatography in the presence of 10 μM compound ‘4905. Monomeric peak fractions were concentrated and used for G protein complex formation. GαoGβ1γ2 heterotrimeric G proteins were expressed in Hi5 insect cells (Expression Systems, 94-002S) and purified using Ni2+ affinity following detergent solubilization and dephosphorylation. The final α2AAR–GαoGβ1γ2–scFv16 complex72 was assembled in the presence of ‘4905 and purified by size-exclusion chromatography. Cryo-EM grids were prepared using UltrAufoil R1.2/1.3 300-mesh grids and vitrified in liquid ethane. Data were collected on a Titan Krios G3 electron microscope equipped with a K3 direct electron detector. Image processing was performed using cryoSPARC, yielding a final reconstruction at approximately 2.8 Å resolution. Model building and refinement were performed with Protein Data Bank (PDB) 7EJ8 as a starting model using ChimeraX73, Phenix74 and Coot75. The final structures have been deposited in the PDB with accession codes 9PLO and 9PLN. Structural representations of protein–ligand complexes were generated using the PyMOL Molecular Graphics System v2.5.5 (Schrödinger).
Mac1 purification and crystallization
Wild-type Mac1 protein (P43 construct, residues 3–169) was expressed in Escherichia coli BL21(DE3) as an N-terminal His6-tagged construct and purified by Ni2+-affinity chromatography71. The His-tag was cleaved with TEV protease, followed by size-exclusion chromatography (Superdex 75) in 20 mM Tris-HCl (pH 7.5), 150 mM NaCl and 1 mM dithiothreitol. Purified protein was concentrated to 40 mg ml−1 for crystallization and stored at −80 °C.
Crystals were obtained by sitting-drop vapour diffusion in 28% PEG 3000 and 100 mM CHES (pH 9.5). Compounds (100 mM in DMSO) were added to crystal drops using an Echo 650 acoustic dispenser76 to a final concentration of 10 mM. Crystals were incubated at room temperature for 2–4 h and vitrified in liquid nitrogen without additional cryoprotection. X-ray diffraction data were collected at 100 K at the Advanced Light Source (beamline 8.3.1) using an X-ray wavelength of 0.88557 Å, and diffraction data were processed using XDS77 and Aimless78. Structures were determined to resolutions ranging from 0.97 to 1.02 Å. Ligands with low occupancy or conformational disorder were modelled using PanDDA79 and Coot75 and refined using phenix.refine80 as previously described81. The structure of ‘3184 was determined using a racemic preparation of the compound (‘9037). X-ray data collection and refinement statistics are shown in Extended Data Tables 1–4, as are the 32 PDB IDs. Structural representations of protein–ligand complexes were generated using the PyMOL Molecular Graphics System v2.5.5 (Schrödinger).
Microsomal stability
Microsomal stability of compounds was evaluated using pooled mouse liver microsomes (M3000/lot #2010026, XenoTech) to estimate their metabolic stability and predict hepatic clearance. Each compound was incubated at 2 μM in a reaction mixture containing 0.42 mg ml−1 microsomal protein, phosphate buffer (100 mM, pH 7.4), MgCl2 (3.3 mM), NADPH (3 mM), glucose-6-phosphate (5.3 mM) and glucose-6-phosphate dehydrogenase (0.67 units per millilitre). Reactions were conducted at 37 °C in 96-well plates with shaking at 100 rpm. Samples were collected at five time points (0, 7, 15, 25 and 40 min), and reactions were quenched by adding five volumes of acetonitrile containing an internal standard. After centrifugation at 5,500 rpm for 5 min, the supernatants were analysed via high-performance liquid chromatography–tandem mass spectrometry (HPLC–MS/MS). The elimination rate constant (kel), half-life (t1/2) and intrinsic clearance (Clint) were calculated by plotting the natural logarithm of the remaining parent compound versus time. Stability was compared with reference standards such as imipramine and propranolol.
Plasma protein binding
Plasma protein binding was measured using equilibrium dialysis with a 14-kDa molecular weight cut-off membrane in a 96-well HTD96b dialyser. Mouse plasma containing 1 μM test compound (0.005% DMSO and 1% acetonitrile) was placed in one chamber, and phosphate-buffered saline (PBS, pH 7.4) in the opposing chamber. The assembled plates were incubated at 37 °C with 5% CO2 and approximately 95% humidity, shaking at 250 rpm for 5 h to reach equilibrium. After incubation, aliquots from each chamber were mixed with equal volumes of the blank opposite matrix and processed with acetonitrile containing internal standard. Supernatants obtained after centrifugation were analysed by HPLC–MS/MS. The percentage of compound bound to plasma proteins was calculated using the peak area ratio in buffer to plasma compartments. Recovery and stability standards were included to ensure accuracy and reliability. Verapamil served as a reference control. Most compounds showed moderate binding, typically ranging between 75% and 85%.
Plasma stability
Plasma stability was assessed in non-sterile mouse plasma (Li-heparin treated) at 1 μM concentration (final DMSO content was 0.005%). Incubations were carried out in aliquots of 60 μl each (two per time point) at 37 °C under 5% CO2 and high humidity (approximately 95%). The reactions were quenched with 240 µl of 90% acetonitrile containing an internal standard at 0, 20, 40, 60 and 120 min, followed by centrifugation at 6,000 rpm for 5 min. Supernatants were analysed via HPLC–MS/MS to determine the percentage of parent compound remaining at each time point. Data were plotted to calculate compound half-lives (t1/2). Reference compounds, verapamil and propantheline, were used as high and low stability controls, respectively. This assay is critical for identifying compounds susceptible to degradation by plasma esterases or hydrolytic enzymes, helping inform PK optimization strategies during lead selection.
Thermodynamic solubility
Aqueous thermodynamic solubility was determined in PBS (pH 7.4) for 247 compounds using a shake-flask method followed by UV absorbance quantification. Dry powder compounds were dissolved in PBS to a theoretical concentration of 4 mM and incubated in duplicate at 25 °C for 4 h and 24 h with shaking. After incubation, samples were filtered using HTS 96-well filter plates. The filtrates were diluted twofold in acetonitrile with 4% DMSO for UV analysis. The incubation samples for charged molecules were additionally diluted tenfold with 50% acetonitrile–PBS with 2% final DMSO. Calibration curves (0–200 μM) were prepared in 50% acetonitrile–PBS (2% final DMSO). Absorbance was measured between 230 and 550 nm using a SpectraMax Plus microplate reader. Compound-specific absorbance maxima were used to calculate concentration using SoftMax Pro and Excel. The assay dynamic range was approximately 2–400 μM (about 20–4,000 μM for charged molecules), with values near the upper limit treated as semi-quantitative. Ondansetron was used as a reference compound. This method reflects equilibrium solubility under physiologically relevant conditions and helps to rank compounds for formulation feasibility.
PAMPA-BBB
Passive blood–brain barrier (BBB) permeability was estimated using PAMPA (PAMPA-BBB) with a phospholipid-coated membrane simulating the brain endothelium. Test compounds (50 μM in Prisma HT buffer, pH 7.4, with 0.5% DMSO) were added to donor wells, whereas brain sink buffer was added to the acceptor wells. The donor and acceptor chambers were separated by a 0.45-µm filter membrane coated with brain polar lipids. Plates were incubated without agitation at room temperature for 4 h. Post-incubation, samples from both chambers, as well as a standard solution, were diluted with acetonitrile containing an internal standard. Apparent permeability coefficients (log Papp) were calculated based on peak area ratio. Clozapine and chlorpromazine were used as high-permeability controls, whereas ranitidine represented low permeability. The assay provides a high-throughput, non-cell-based method for estimating central nervous system exposure potential.
Behavioural analyses
Male C57BL/6J mice 7–8 weeks of age were obtained from The Jackson Laboratory (JAX strain #000664) and housed five per cage under a standard 12–12-h light–dark cycle at 20.7 °C (69.3 °F) and 58% humidity. All animal procedures were approved by the University of California, San Francisco Institutional Animal Care and Use Committee (protocol #AN208219). Animals were randomly assigned to treatment and control groups. For behavioural experiments, mice were initially placed together in a cage and allowed to move freely for a few minutes; each mouse was then randomly selected, injected with compound or vehicle, and placed in a separate cylinder before testing. The experimenter assessing behaviour was different from the experimenter administering injections and placing mice into the cylinders; experimenters were therefore blinded to treatment. Behavioural experiments were conducted in two independent cohorts. Mice were habituated individually in Plexiglas enclosures for 1 h before testing. Compounds were administered subcutaneously 30 min before behavioural assessment, and, where applicable, the α2AAR antagonist atipamezole (2 mg kg−1, intraperitoneally) was given 15 min before compound injection. Tail-flick latency was measured by immersing the distal third of the tail in a 50 °C water bath and recording the withdrawal time. For the neuropathic pain model, SNI was performed under isoflurane anaesthesia by ligating and transecting two of the three branches of the sciatic nerve, sparing the sural nerve. Mechanical thresholds were assessed 7–14 days post-surgery using von Frey filaments and the up–down method, and values were normalized to the baseline of each animal. Thermal nociception was evaluated using a 55 °C hotplate, and the latency to nocifensive behaviour (paw lick or jump) was recorded with a cut-off of 45 s to prevent tissue damage. Behavioural data are summarized as group size (n), mean and s.d. (Supplementary Table 13). Statistical analyses were performed in GraphPad Prism. For the SNI experiment (Fig. 5f), we used two-way ANOVA followed by Tukey’s multiple-comparisons test. The hotplate assay (Fig. 5g) was analysed using the Friedman test followed by Dunn’s multiple-comparisons test. For the tail-flick assays, the 4905 dose–response (Fig. 5c) was analysed by two-way ANOVA followed by Dunnett’s multiple-comparisons test against vehicle (control), whereas the PS75 and 3629 dose groups (Fig. 5d,e) were analysed using the Kruskal–Wallis test followed by Dunn’s multiple-comparisons test against vehicle. The Kruskal–Wallis and Friedman tests were nonparametric; Prism calculated two-sided, multiplicity-adjusted P values for the Dunn comparisons (Supplementary Table 13). For sample size, we did not perform formal a priori power calculations. Instead, group sizes were guided by our previous experience with these assays and by precedent in the literature. This approach may limit sensitivity to small effects. Raw data for the animal assays underlying Fig. 5c–g are provided in the Source Data file.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
