Atomic structure of human TOM core complex

The translocase of the outer mitochondrial membrane (TOM) complex is the main entry gate for mitochondrial precursor proteins synthesized on cytosolic ribosomes. Here we report the single-particle cryo-electron microscopy (cryo-EM) structure of the dimeric human TOM core complex (TOM-CC). Two Tom40 β-barrel proteins, connected by two Tom22 receptor subunits and one phospholipid, form the protein-conducting channels. The small Tom proteins Tom5, Tom6, and Tom7 surround the channel and have notable configurations. The distinct electrostatic features of the complex, including the pronounced negative interior and the positive regions at the periphery and center of the dimer on the intermembrane space (IMS) side, provide insight into the preprotein translocation mechanism. Further, two dimeric TOM complexes may associate to form tetramer in the shape of a parallelogram, offering a potential explanation into the unusual structural features of Tom subunits and a new perspective of viewing the import of mitochondrial proteins.


Cryo-electron microscopy
4-μL aliquots of digitonin-solubilized TOM-CC at a concentration of 4.5 mg/ml were applied to discharged 400-mesh Quantifoil R1.2/1.3 grids (Quantifoil, Micro Tools GmbH, Germany). Grids were blotted for 3.5 sec and plunged into liquid ethane using an FEI Mark IV Vitrobot operated at 4 °C and 100% humidity. Micrographs were collected on a Titan Krios microscope operated at a voltage of 300 kV with a Gatan K2 Summit direct electron detector. Images were recorded using a 300 kV Titan Krios electron microscope with serialEM software (Thermo Fisher) at a nominal magnification of 130,000 × using super-resolution mode. The pixel size was 1.08 Å/pixel and the defocus range were set from -1.3 μm to -2.3 μm. The total dose rate on the detector was about 50 e/ Å2 with a total exposure time of 8 s. Each micrograph stack contains 32 frames. Four batches of data were collected with two cryo-samples, obtaining 5,908 micrographs in total.
Each micrograph was corrected for sub-region motion correction and dose weighting using UCSF MotionCor2 2 . Gctf was used to determine the contrast transfer function (CTF) parameter and produce the CTF power spectrum on basis of summed micrographs from MotionCor2 for all micrographs 3 . The 5,443 CTF-corrected cryo-EM images were manually selection. Particles were auto-picked on micrographs with dose-weighting using RELION 4 . Briefly, about 1,000 particles were manually picked from a subset of images and extracted in a box size of 200 pixels and a mask diameter of 200 Å. Extracted particles were subjected to 2D classification requesting 10 classes, 8 classes of which showed representative views and were selected as templates for automated particle picking. The resulting 2D averages were served as the templates for particle auto-picking, 1,638,577 particles picked from 5,443 images. For the dataset, particle selection, 2D and 3D classifications were performed on a binned dataset with a pixel size of 5.4 Å using RELION, which were then manually inspected to exclude noise and other bad particles. Two rounds of 2D classification requesting 100 classes were resulted in 823,450 particles, and two rounds of 3D classification which contained 408,208 particles and 30,916 dimeric TOM-CC like particles. An initial model was generated from 20,000 best particles using cryoSPARC ab initio reconstruction requesting five classes with C1 symmetry 5 . A total of 347,796 particles from these 3D classes was re-extracted to the original pixel size of 1.08 Å, and classified into five classes with C1 symmetry using a reference model generated from cryoSPARC, which had been low-pass filters to 20 Å 5 . Two rounds of 3D classification and the most populated class containing 159,369 particles was subjected to further 3D auto-refinement with C2 symmetry. The refinement resulted in an overall structure at a resolution of 3.91 Å, which allowed initial model building. To further improve the resolution, we performed CTF refinement, which yielded a map at 3.43 Å resolution.
The 30,916 tetrameric TOM complex like particles performed two rounds 2D classification and one round 3D classification which obtained 11,184 good particles. After re-extracted to the original pixel size of 1.08 Å, 11,184 particles were subjected to further 3D auto-refinement with C2 symmetry. The global resolution for tetrameric TOM complex was 8.53 Å. All reported resolutions are based on the gold-standard Fourier Shell Correlation (FSC) = 0.143 criteria 6 , and the final FSC curve was corrected for the effect of a soft mask using high-resolution noise substitution. Final density maps were sharpened by B-factors calculated with the RELION post-processing program. All reported resolutions are based on the gold-standard FSC = 0.143 criteria, and the final FSC curve was corrected for the effect of a soft mask using high-resolution noise substitution. The local resolution map was calculated using ResMap 7 (Supplementary Fig. S1d-e, 2, Table S1).

Quantification and Statistical Analyses
After refinement, map CC between model and EM map was 0.66, indicative of a reasonable fit at the present resolution. The resulting model was also used to calculate a model-map FSC curve, which agreed well with the gold-standard FSCs generated during the RELION refinement. The final model has good stereochemistry, as evaluated using MolProbity 8 .

Model Building, Refinement and Validation
Atomic models of the TOM-CC subunits were predicted and modeled using the webserver (http://www.sbg.bio.ic.ac.uk/phyre2) of Phyre2 9 . We assigned homology models of the TOM-CC subunits, Tom40, Tom 22, Tom7, Tom6 and Tom5 into our density maps. The models were further optimized by Coot 10 . For cross-validation against overfitting, we randomly displaced the atom positions of the final model by up to a maximum of 0.5 Å, and refined against the half map generated by RELION 3D auto-refine procedure, resulting a model named Test. Then we calculated the FSC curve of both half map against the model Test and compared with the FSC curve of the final model against the summed map generated by RELION 3D auto-refine procedure. To build the human tetrameric TOM complex models, we fitted the density maps of TOM-CC into the 8.53 Å tetramer map and combined these part maps into a whole map in UCSF Chimera. Based on the combined map, the human tetramer TOM complex model was built. And then, the model with the ligands were subjected to global refinement and minimization in real_space_refinement using PHENIX. All ligands and phospholipids models were generated using elbow 11 module in PHENIX 12 by their geometric constraints. The phospholipid was docked into densities and refined in COOT. All the figures were created in PyMOL 13 , Chimera and UCSF Chimera(X) 14 , quantitative electrostatics were calculated using APBS 15 , and conservation levels were generated with ConSurf 16 , sequences alignment of Tom subunits were generated by the ESPript 17 , the structure predictions of human & yeast Tom subunits are based on the Predict Secondary Structure (PSIPRED) 18 .

Cell Culture and Transient knockdown of Tom subunits
Hela (Thermo Fisher Scientific) cells were grown in DMEM media (Corning) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific) and cultured in a 37°C, 5% CO2 incubator. 4.5x10^5

Western Blotting
Harvested Hela cells were lysed by 1% digitonin (1% digitonin, 25mM Tris pH=7.4, 150mM NaCl) at 4℃ for 30min. After centrifugation at 13,000xg, 4℃ for 15min, the supernatant was collected for Blue native PAGE (BN-PAGE) with the acrylamide concentration from 3% to 12%. The separated proteins were electro-transferred to PVDF membranes (Millipore) at 300 mA for 1h. The membranes were incubated with acetic acid for 8min, followed with incubation with methyl alcohol for 5min for 3 times.

Supplementary Fig. S4 | Sequence comparison of TOM-CC subunits between different species.
The threshold value of the alignment is 0.7, which means if the similarity score assigned to a column is greater than this value, residues are considered as highly similar and are colored in red and framed in blue. Secondary structure based on human Tom subunits are indicated. The alignments were generated by the ESPript.  Cross-linked peptides of Tom subunits and Tim subunits.