Cryo-EM analysis of the PaExoY-F-α-actin complex

Rabbit muscle F-actin was prepared as described previously³⁰. In brief, the freshly thawed protein was spun down using the TLA-55 rotor for 30 min at 150,000 × g at 4 °C, and the G-actin-containing supernatant was collected. Then, the protein was polymerized by incubation in the buffer containing 120 mM KCl, 20 mM Tris pH 8, 2 mM MgCl₂, 1 mM DTT, and 1 mM ATP (F-buffer) in the presence of a twofold molar excess of phalloidin for 30 min at room temperature and further overnight at 4 °C. The next day, the actin filaments were pelleted using the same TLA-55 rotor for 30 min at 150,000 × g at 4 °C and resuspended in F-buffer. Fifteen minutes before plunging, F-actin was diluted to 2 µM, mixed with 4 µM of PaExoY, 2 mM 3′-deoxyguanosine-5′-triphosphate (3′-dGTP, Jena Bioscience) and 4 mM MgCl₂. Shortly before plunging, Tween-20 was added to the sample to a final concentration of 0.02% (w/v) to improve the ice quality. Plunging was performed using the Vitrobot Mark IV system (Thermo Fisher Scientific) at 13 °C and 100% humidity. 3 µl of sample were applied onto a freshly glow-discharged copper R2/1 300 mesh grid (Quantifoil), blotted for 8 s on both sides with blotting force −20 and plunge-frozen in liquid ethane.

The dataset was collected using a Krios Titan transmission electron microscope (Thermo Fisher Scientific) equipped with an XFEG at 300 kV using the automated data-collection software EPU, version 2.7 (Thermo Fisher Scientific). Five images per hole with a defocus range of −0.4 to −3.5 µm were collected with the Falcon III detector (Thermo Fisher Scientific) operated in linear mode. Image stacks with 40 frames were collected with a total exposure time of 1.5 s and total dose of 93 e⁻/Å-. 12437 images were acquired and 8663 of them were used for processing. Motion correction and CTF estimation were performed in CTFFIND³⁵, version 4.1.1, and MotionCorr2³⁶, version 1.1.0, during image acquisition with TranSPHIRE³⁷, version 1.4.28. Filament picking was performed using crYOLO³⁸ version 1.5. On the next step, 2.25 million helical segments were classified in 2D using ISAC³⁹ in Sphire version 1.3 to remove erroneous picks. The remaining 1.85 million particles were used in the first 3D refinement in Meridien⁴⁰ in Sphire version 1.3 with 25 Å low-pass filtered F-actin map as the initial model and with a spherical mask with a diameter of 308 Å, followed by a local 3D refinement with a wide mask of the shape of the PaExoY-F-actin complex. On the next step, per-particle CTF-refinement was performed using 3D refinement parameters and the 3D reconstruction in Sphire, version 1.3, followed by removal of the segments with defocus value lower than −3 µm and recentering of the remaining 1.71 million particles according to the parameters calculated in the previous 3D refinement. Following the third round of 3D refinement in Sphire, Bayesian polishing and 3D classification were performed in Relion version 3⁴¹. The latter step was performed with a mask covering three actin and one PaExoY subunit in order to remove particles that do not possess bound PaExoY in the particular position. Indeed, 176067 segments without PaExoY density were removed and the remaining 1.54 million were introduced into the final round of 3D refinement with a tight mask covering actin and PaExoY subunits. The final reconstruction map was postprocessed using 3D local filter based on the local resolution of the map in Sphire. Maps presented on Fig. 1a and Supplementary Movie ² were postprocessed using DeepEMhancer version 1.0⁴². The overall processing, FSC curves and local resolution maps are available in Supplementary Fig. ¹.

To build a model of the PaExoY-F-actin complex, we performed flexible fitting of the PaExoY partial crystal structure (5XNW²²) into the EM density using iMODFIT Chimera plugin⁴³, version 1.2. Then, we build the remaining 106 amino acids using Rosetta software version 3^44,45 and added 3′-dGTP, which was modeled with eLBOW⁴⁶. F-actin, ADP-_Pi and Phalloidin were adapted from PDB 7AD9³⁰. The model was further refined using ISOLDE⁴⁷, version 1.0B4, and Phenix⁴⁸, version 1.17. Schematic representation of the secondary structure elements of PaExoY for Fig. 1c was calculated in UCSF Chimera, version 1.14, with H-bond energy cutoff −0.45 kcal/mol.

The Buried surface area of the PaExoY-F-actin interface and distances between ligands and neighboring atoms (Supplementary Fig. ²ⁱ) were calculated using PDBsum⁴⁹.