Structural insights into tecovirimat antiviral activity and poxvirus resistance

Abstract

Mpox is a zoonotic disease endemic to Central and West Africa. Since 2022, two human-adapted monkeypox virus (MPXV) strains have caused large outbreaks outside these regions. Tecovirimat is the most widely used drug to treat mpox. It blocks viral egress by targeting the viral phospholipase F13; however, the structural details are unknown, and mutations in the F13 gene can result in resistance against tecovirimat, raising public health concerns. Here we report the structure of an F13 homodimer using X-ray crystallography, both alone (2.1 Å) and in complex with tecovirimat (2.6 Å). Combined with molecular dynamics simulations and dimerization assays, we show that tecovirimat acts as a molecular glue that promotes dimerization of the phospholipase. Tecovirimat resistance mutations identified in clinical MPXV isolates map to the F13 dimer interface and prevent drug-induced dimerization in solution and in cells. These findings explain how tecovirimat works, allow for better monitoring of resistant MPXV strains and pave the way for developing more potent and resilient therapeutics.

Fig. 1. F13 forms a homodimer that can be inserted into a membrane’s surface.

a, Schematic representation of the replication cycle of OPXVs. Mature viruses enter the cell, fusing their membrane (in blue) with the cellular one. After DNA replication, immature particles are formed (IV, membrane in red), which give rise to intracellular mature virus particles in the cytoplasm of the infected cell. Mature viruses (MV) can either be released by lysis or wrapping. In the latter, mature viruses acquire two additional membranes (WV, in orange) from the Golgi apparatus or endosomal vesicles to form wrapped virions, fuse the outermost with the plasma membrane and release enveloped viruses (EV). Tecovirimat blocks wrapping, as indicated. The scheme, adapted from Fig. 1 in ref. , was created with BioRender.com. b, Crystal structure of the sF13 homodimer represented in cartoon. One protomer is coloured blue and the other green. The N termini and the MIR are indicated on one protomer, and the phospholipase active site is indicated on the other. Bottom panels provide close-up views of the two regions forming the dimer interface, indicated by coloured rectangles in the upper panel. All single escape mutants identified to date are shown as spheres, coloured according to their potency, reported as IC50 fold change. c, Side view of the F13 homodimer interacting with a lipid membrane that mimics Golgi membrane composition, as observed from molecular dynamics simulations. For clarity, water molecules and lipids in the foreground of the membrane are not shown. sF13 chains are coloured as in b, with palmitoylated cysteines and hydrophobic residues in the MIR and N termini depicted as sticks. The bottom panel provides close-up views to show lipid–protein interactions, with the protein residues involved in the interaction depicted as sticks and labelled. Protein carbons are coloured according to the chain, membrane carbons in white. Nitrogen, oxygen, sulfur and phosphate atoms are coloured blue, red, yellow and orange, respectively.

Fig. 2. Tecovirimat binding site.

a, Cubic crystals used to obtain the structure of the sF13/tecovirimat and sF13/IMCBH complexes. b, Crystal structure of the sF13/tecovirimat complex. Left: a Fo-Fc omit map contoured at 3σ showing the density found at the dimer interface in the soaked crystal with the tecovirimat molecule modelled. Centre and right: orthogonal views of the dimerization interface with the tecovirimat molecule modelled and the residues contacting the drug represented as sticks and labelled. sF13 chains are coloured as in Fig. 1. c, Crystal structure of the sF13/IMCBH complex. Left: as in b, an omit map showing the electron density at the dimer interface in the soaked crystal. Centre and right: as in b, orthogonal views showing the sF13/IMCBH contacts.

Fig. 3. Tecovirimat induces sF13 dimerization in solution.

a, AUC analysis of sF13 without tecovirimat (brown line) and with 10 µM tecovirimat (blue line). Experimentally derived sedimentation coefficient values (Svedberg units (S)) are shown above each peak. c(s), the concentration of protein with sedimentation coefficient s. b, Experimental SAXS profile (green dots) and theoretical profiles (dashed lines) calculated using CRYSOL for one monomer of sF13 (blue dashed line) and the dimer shown in Fig. 1 (pink dashed line). I(q), scattering intensity in function of the scattering vector q; q (Å⁻¹), scattering vector; Å, Angstrom; Χ², chi-squared test. c, Orthogonal views of a representative dummy atom model (green) reconstructed from SAXS data. For comparison, we have included below a model of the crystallographic sF13 dimer with an outline of the dummy model. d, Dose–response curve used to estimate tecovirimat effect in solution. The y axis represents the proportion of dimers in a dilute solution of F13 measured by mass photometry. The x axis represents the concentration of drug (tecovirimat or IMCBH) present in the solution. The EC50 values were determined from a dose–response curve fitted using GraphPad Prism. Data are mean ± s.d. of three independent experiments (n = 3). e, Tecovirimat (blue line) and IMCBH (orange) inhibit plaque formation of MPXV. Vero cells were infected with MPXV clade IIb and treated with the indicated concentrations of tecovirimat or IMCBH. Plaque inhibition is expressed as a percentage, normalized to control conditions. Data are mean ± s.d. of triplicate wells from five independent experiments for tecovirimat (n = 15) and two independent experiments for IMCBH (n = 6). Source data

Fig. 4. Escape mutants identified in mpox patients prevent tecovirimat-induced dimerization.

a, Mass-photometry-based dose–response curve showing tecovirimat activity against different escape mutants, as indicated. Data are mean ± s.d. of three independent experiments (n = 3). b, AUC analysis of sF13^A295E (left panel) and sF13^4MUT (right panel) without tecovirimat (brown line) and with tecovirimat (black line). Experimentally derived sedimentation coefficient (S) values are shown above each peak. c,d, Orthogonal views showing the dimer interface of sF13^A295E (cyan, c) and sF13^A295E/tecovirimat (green, d) superimposed on sF13^WT (orange). Residues E295, R291, Y285 and N300 are represented as sticks and labelled. Polar contacts are indicated with dashed lines. Source data

Fig. 5. Tecovirimat induces F13 dimerization in cells.

a, Schematic model representing the PLA experiment. F13 protomers are coloured green and cyan with the approximate location of the flag tag indicated with a blue sphere. The three steps of the assay, dimerization, ligation and amplification, are indicated. b, Upper panel: representative fluorescence microscopy images with the nuclei coloured in blue and the PLA signal in red. Scale bar, 100 µm. Lower panel: quantification of the PLA signal as the average area of PLA fluorescence per cell; 7,000 to 12,000 cells were analysed per data point. Data are mean ± s.d. of two independent experiments performed in triplicate (n = 6). For statistical analysis, two-way analysis of variance was used. NS, non-significant; ****P < 0.0001; **P = 0.0087. NT, non-transfected. Source data

Fig. 6. Structure-based escape mutations do not generate viable viruses.

a, Close view of the dimerization interface across the twofold axis showing the designed mutations S292F, S292K and L296Y and the mutation identified in VARV, R291E. The circle indicates the localization of the tecovirimat-binding site. b, Mass-photometry-based dose–response curve showing tecovirimat activity against different mutants, as indicated. Data are mean ± s.d. of three independent experiments (n = 3). c, Viral titres in p.f.u. ml⁻¹ (left panel) and plaque size (right panel) in the presence (+) and absence (−) of 10 µM tecovirimat calculated from plaque assays. Each bar represents the means ± s.d. from two independent experiments (n = 2). The limit of detection (10² p.f.u. ml⁻¹) is indicated with a dashed line. Source data

Extended Data Fig. 1. Multiple sequence alignment.

Multiple sequence alignment from six representative OPXVs. The secondary elements are indicated at the top. Strictly conserved residues are highlighted in red. The palmitoylated cysteines are shown in blue and the membrane interacting region (MIR) framed and labeled. Tecovirimat contacts and positions of escape mutants are marked with black circles and triangles under the alignment, as indicated. The accession codes for the F13 proteins used in the alignment are: Borealpox virus (BOPX, QED21148.1), Camelpox virus (CMLP, A0A0K1LD56), Variola virus (VARV, AAA60785.1), Vaccinia virus (VACV, P04021), Monkeypox virus (MPXV, YP_010377040.1), Cowpox virus (CPXV, CAD90601.1). the alignment was performed using clustal omega ⁸⁹ and the figure prepared with ESPRIPT ⁹⁰.

Extended Data Fig. 2. Contacts at the dimer interface.

a) The left panel shows the crystal structure of the sF13 homodimer (PDB: 9FHS) represented in cartoon form. One protomer is colored blue, and the other is green. The 2-fold axis is indicated by a black line. The right panels provide two close- up views of the dimer interface, as indicated in the left panel. The purple and yellow arrows indicate a top view or a bottom view of the dimer interface, respectively, as shown in the left panel. The main residues contributing to the dimer interface (identified using the PDBePISA server) are depicted as sticks and labeled (green or blue). Polar contacts are shown as dashed black lines. b) Similar to A, the left panel shows the crystal structure of the PLD3 homodimer (PDB: 8V05), with protomers colored green and blue. The right panels are close-up views of the dimer interface. c) Similar to A, the left panel shows the crystal structure of the PLD4 homodimer (PDB: 8V08), and the right panels are close-up views of the dimer interface.

Extended Data Fig. 3. Cluster analysis of the 45 input structures for free energy perturbation simulations.

Analysis of putative binding poses after equilibrating the 15 selected poses (x-label, first number) by three independent MD replicas (x-label, second number, see Methods). To reveal reoccurring protein-ligand interaction networks, a cluster analysis was performed. The x-axis represents the pose identification, and the y-axis indicates the distance between each pose in terms of their protein-ligand interaction network. The lowest energy pose of each cluster is highlighted, with its protein–ligand interaction profile and corresponding absolute binding free energy. The analysis demonstrates (together with Supplementary Table 4) that different protein-ligand interaction networks yield similar absolute binding free energies. The pose with the strongest binding affinity of -25.6 kcal/mol was used for further refinement against the crystallographic data, as reported in the PDB file. The Python library ProLIF was used to generate the protein-ligand interaction network.

Extended Data Fig. 4. Validation of the tecovirimat conformation using seven structurally similar ligands.

Eight ligands, including tecovirimat, were aligned to the proposed tecovirimat binding pose, and the absolute binding free energy for each ligand binding to the dimer was calculated (see Methods). Pearson, Kendall, and Spearman correlation coefficients are reported alongside root mean square error (RMSE), mean signed error (MSE), and mean unsigned error (MUE). Ligands are identified in the validation plot by numbers taken from. Top right: 3D alignment of the eight molecules within the binding pocket. Bottom: chemical structures of the ligands with the corresponding EC50 values. The alignment was rendered with PyMOL⁹¹ and chemical structures were drawn with RDKit⁹².

Extended Data Fig. 5. SAXS analysis of sF13/tecovirimat.

a) Guinier plot showing the experimental scattering curve of the F13/tecovirimat complex in green, and the fitted curve used to generate the pair distance distribution in orange. b) Pair distance distribution function calculated using GNOM used to obtain Dmax and Rg values. c) Dimensionless (normalized) Kratky plot showing the characteristic shape of a well-folded protein. Source data

Extended Data Fig. 6. Mass photometry (MP) assay.

Each row shows the pipeline used to determine EC50 values for sF13 wild-type and the different mutants, as indicated. The left column displays the mass distribution for sF13 at 25 nM without tecovirimat. The middle column shows mass distribution curves for sF13 with increasing drug concentrations, as indicated, highlighting in red the region used to calculate the percentage of dimers for the dose-response curve in the right column. The EC50 is derived from this dose- response curve. For clarity, the middle column presents a single representative experiment per drug concentration; however, the dose-response curve is based on the mean and standard deviation from three repeated experiments.

Extended Data Fig. 7. Image quantification methodology and percentage of F13- FLAG transfected cells.

Top panels) Quantification of the PLA signal area per cell. To measure the extent F13 dimerization, first, the number of Hoechst positive nuclei were automatically counted. Second, the PLA positive area was delimited and measured. Finally, the total PLA area was divided by the total number of nuclei. Scale bar: 50 μm. Middle panels) For percentage of transfection, total number of nuclei were automatically counted and delimited, then the total number of F13-FLAG positive nuclei were counted. Percentage of transfected cells was calculated by dividing the number of positive nuclei by the total number of nuclei. Scale bar: 50 μm. Bottom panels) Quantification of the number of F13-FLAG positive cells for the indicated treatments. Detection with anti-Mouse and anti- rabbit secondary antibodies is indicated. 7000 to 12000 cells were analysed per data point. Data are mean±sd of two independent experiments performed in triplicat (n=6). Statistical analysis: Two-Way ANOVA. ns: non-significant. *p=0.0228, **p=0.0033, ****p<0.0001. Source data

Extended Data Fig. 8. Proteins and crystal analysis.

a) SDS-PAGE of the proteins used in this manuscript. Molecular weight markers are shown in the left lane. In the other lanes, 1 μg of sF13 wild-type and mutants have been loaded, as indicated. For the sake of clarity, we have prepared a gel with all proteins at the same time, side by side. All of them have been analyzed at least 3 times on SDS-PAGE with identical results. b) Crystal structure of the sF13/tecovirimat complex processed in different space groups, as indicated. 2Fo-Fc maps contoured at 1σ are shown in blue, Fo-Fc maps contoured at +3σ and -3σ are shown in green and red, respectively. In the cubic space group (F432), a 2-fold symmetry axis passes through the center of the tecovirimat molecule, so the density represents two tecovirimat molecules with 50% occupancy each. Only one of the molecules is shown. In the central panel, we display the electron density resulting from refining the structure in P1, using a single orientation for tecovirimat (100% occupancy). Similarly, in the right panel, we process the data in P1 but refine the structure using two orientations with 50% occupancy each. Source data

Extended Data Fig. 9. MD simulations of the F13 dimer on a lipid membrane.

(a) Contact map showing interactions between the two monomers, calculated from the X-ray structure. (b) Contact map showing interactions between the two monomers, calculated from MD simulations by concatenating the last 300 ns from five repeats. The contact map highlights monomer-monomer interactions within 5 Å.

Extended Data Fig. 10. Protein-ligand interaction network analysis of the last 10 ns of the free simulation during the equilibration phase for pose 6-3.

a) Tanimoto similarity matrix representing ligand-protein interactions across each frame of the MD trajectory. Values range from 0 to 1, where 1 indicates the highest similarity and 0 indicates the lowest. b) Barcode plot of interactions. Each horizontal line represents the presence of the corresponding interaction at a specific frame of the MD trajectory.