Preparation of high-quality graphene filmsGraphene films with large single-crystal domains were prepared via chemical vapor deposition (CVD) method. 27-μm-thick Cu foils (Kunshan Luzhifa Electron Technology Co., Ltd., China) were placed in a homemade low-pressure CVD system, which is equipped with a 6-inch quartz tube and three heating zones. An asynchronous heating process and a temperature gradient were applied to the Cu foils, which promoted the enlargement of Cu single crystals29. Then the samples were annealed for 30 min in a mixture of Ar and O2 (PAr = 1000 Pa, PO2 = 0.4 Pa) and 10 min in a mixture of Ar and H2 (PAr = 500 Pa, PH2 = 500 Pa), respectively. The graphene growth proceeded under a gas mixture of H2 (PH2 = 500 Pa), Ar (PAr = 500 Pa), and CH4 (PCH4 = 1.6 Pa) for 90 min. The CVD chamber was cooled down to room temperature after growth procedure, during which the same atmosphere and partial pressures were maintained. The high-quality graphene films are also commercially available in Beijing Graphene Institute (BGI) company (http://www.bgi-graphene.com/type/100).Preparation of free-standing GSAMs superstructurePlasma treatment of Cu foilTo remove graphene on the back side of Cu foil, we put Cu foil into a reactive ion etcher (Pico SLS, Diener) with an airflow of 10 sccm. Then the air plasma was generated at a controlled power of 150 W. After a 3-min treatment, the backside graphene was fully removed.Self-assembly of stearic acid monolayersStearic acid was dissolved and diluted in isopropanol (IPA) to reach the mass fraction of 0.005%. Then we used a capillary to drip the solution on graphene to fully cover its surface. Normally, each drop volume of stearic acid/IPA solution we use was about 10 μL, which could spread over an area of about 1.5 cm × 1.5 cm. After the evaporation of IPA, stearic acid monolayer was self-assembled on graphene surface.EtchingThe Cu/graphene/SAM composite was placed on the surface of 0.5 M (NH4)2S2O8 aqueous solution to etch the Cu foil away. After etching, the free-standing graphene/SAM superstructure was floated at the air–water interface.Preparation of suspended GSAMs membranesTEM grids (e.g., Quantifoil-R1.2/1.3, 300 mesh) were carefully positioned on the filter paper, which had been submerged into the etching solution in advance. The solution was then gradually pumped away to lower down the solution surface, making the floating GSAMs superstructure deposit onto TEM grids.RinsingGSAM-coated TEM grids were transferred away from etching solution and submerged into deionized water to wash away the inorganic contaminations. The GSAMs grid could be rinsed in the isopropanol solution for 30 s to remove the organic contaminations if needed.DryingThe suspended GSAMs membranes on the grids were air-dried in the cleanroom to avoid extra contaminants after being taken out from the deionized water.Preparation of pristine graphene gridsThe pristine graphene grids are prepared by a polymer-free clean transfer method22,30. Firstly, the commercial holey carbon grids (Quantifoil, Au 300 mesh R1.2/1.3) are placed on the surface of graphene/copper substrate. Then, an isopropanol solution is dropped onto these grids. After the isopropanol evaporates, the holey carbon grids will be tightly attached to the surface of the graphene/copper substrate. Subsequently, 1 M (NH4)2S2O8 aqueous solution is used to etch the copper substrate away, resulting in the transfer of graphene film onto the surface of holey carbon grids. After rinsed in deionized water and isopropanol solution, the clean graphene grids can be obtained.Coverage of GSAMs gridThe GSAMs grid coverage is determined by calculating the ratio of the number of holes with intact membranes to the total number of holes in each grid square. We collected the SEM images of 104 grid squares from the GSAMs grid (Supplementary Fig. 2b), and each grid square has ~430 holes (Supplementary Fig. 2c). By counting the number of holes covered with broken, suspended GSAMs membranes, the GSAMs coverage can be obtained. The coverage statistics based on these 104 grid squares have been shown in the Fig. 2e.Characterization of GSAMs by AFMThe surface roughness of suspended SAMs membranes was characterized by atomic force microscopy (AFM, Bruker Dimension Icon) in tapping mode with a commercial AFM tip (OTESPA-R3, Bruker company). The material, tip radius, spring constant, and resonance frequency of OTESPA-R3 AFM tips are 0.01–0.02 ohm-cm Silicon, ~7 nm, ~26 N/m, and ~300 kHz, respectively. The rate of data collection is 1 line/s for 256 lines. The values of surface roughness can be directly obtained from the AFM image.Density functional theory (DFT) calculationsAll the DFT calculations were performed using the Vienna Ab initio Simulation Package (VASP)31,32. The electron-ion and electron-electron interactions were described by the projector augmented wave (PAW) methods and Perdew-Burke-Ernzerh of type exchange-correlation functions, respectively33,34. The 3 × 1 × 1 and 1 × 4 × 1 k-point meshes were adopted for the system of the SAMs aligned along armchair and zigzag directions, respectively. All the structures were released until the force acting on each atom was less than 0.01 eV/Å, with the energy convergence of 10-4 eV. The plane wave cut-off energy was set as 400 eV. A vacuum layer with a thickness greater than 15 Å was constructed to avoid the interactions between two neighboring images. Besides, all calculations include spin-polarization and DFT-D3 was chosen to correct the vdW interaction between graphene and SAMs35.Characterization of GSAMs by cryo-EMThe aforementioned TEM grid coated by suspended GSAMs was directly inserted into liquid nitrogen (−196 °C), and uploaded into a Thermo Fisher Scientific Titan Krios (300 kV) TEM, which is equipped with a direct electron detector K3 (Gatan, Inc.) and a GIF-Quantum energy filter. The suspended GSAMs membrane was imaged at a nominal magnification of 165,000, with a calibrated pixel size of 0.5141 Å. The total irradiation dose is 50 e-/Å2.Binding energy between the graphene and the SAMsThe periodic models of two surfactant molecules on graphene lattice were built to calculate their binding energies (\({E}_{b}\)) by the following equation,$${E}_{b}=\frac{\left({E}_{{SAM}-{Gr}}-{E}_{{SAM}}-{E}_{{Gr}}\right)}{{N}_{{Gr}}}$$
(1)
in which \({E}_{{SAM}-{Gr}}\), \({E}_{{SAM}}\), \({E}_{{Gr}}\), and \({N}_{{Gr}}\) are the total energy of SAMs adsorbed on monolayer graphene, the energy of SAMs, the energy of monolayer graphene and the number of carbon atoms in monolayer graphene, respectively. Finally, the binding energy between SAMs and graphene arranged along the zigzag lattice direction was calculated to be −199.0 mJ/m2.Cryo-EM sample preparationStreptavidin was purchased from New England Biolabs (Catalog: N7021S). To purify the 20S proteasome, its β subunit was His-tagged and expressed in Escherichia coli cells. The protein was then purified using nickel column (GE Healthcare) and stored at 0.6 mg/mL for subsequent usage18. ACE2 was purified from HEK293T cells using a C-terminal six-histidine tag and a nickel affinity column (GE Healthcare). SARS-CoV-2 RBD, tagged with a C-terminal six-histidine tag, was purified from HEK293F cells and mixed with ACE2 at a molar ratio of 1:1.5 to form the ACE2-RBD complex. The Spike protein, tagged with a C-terminal Flag, was purified from HEK293F cells36. All three samples were diluted to 0.25 mg/mL for cryo-EM specimen preparation. CHAPSO (3-(3-cholamidopropyl-dimethylammonio)-2-hydroxy-1-propanesulfonate) was purchased from Sigma-Aldrich (Catalog: C3649-500MG), and DDM (dodecyl-β-D-maltoside) was purchased from Inalco Pharmaceuticals (Catalog: 69227-93-6).To prepare the cryo-EM specimen, 3 μL sample was pipetted onto the glow-discharged (Harrick Plasma, low-power mode for 15 seconds) graphene grid or GSAMs grid, which was then blotted for 0.5 s by filter papers with the force −2 in a Vitrobot (Thermo Fisher Scientific) under 8 °C and 100% humidity. These grids were then directly plunged into liquid ethane and transferred to liquid nitrogen for storage.To prepare the cryo-EM specimen in the presence of surfactants, we added CHAPSO or DDM into the sample solution at final concentrations of 8 mM and 0.17 mM and then applied the sample solution onto glow-discharged EM grids (Quantifoil, 200 mesh, 1.2/1.3). The grids were blotted for 2.5 s with the force −2 in a Vitrobot (Thermo Fisher Scientific) under 8 °C and 100% humidity. After blotting, the grids were frozen by liquid ethane and stored in liquid nitrogen.Single-particle cryo-EM data acquisition and analysisFor 20S proteasome on GSAMs grid, ACE2-RBD and Spike on both GSAMs grid and pristine graphene grid, cryo-EM datasets were automatically collected by EPU software on a Thermo Fisher Scientific Titan Krios G4 (300 kV) TEM equipped with a direct electron detector Falcon 4 (Thermo Fisher Scientific), with the defocus from −1.5 to −1.8 μm and pixel size of 0.86 Å, and we collected 1513, 1445, 3810, 921, and 4352 micrographs for aforementioned samples, respectively. For ACE2 and streptavidin on GSAMs grids, 2448 and 2144 micrographs were collected via a Thermo Fisher Scientific Titan Krios G3i (300 kV) TEM equipped with a direct electron detector K3 (Gatan, Inc.) and GIF-Quantum energy filter by AutoEMation software37, with the defocus from −1.2 to −1.5 μm and pixel size of 0.5191 Å. For ACE2 on graphene grid, 5576 micrographs were collected via a Thermo Fisher Scientific Titan Krios (300 kV) TEM equipped with a direct electron detector K3 (Gatan, Inc.) and GIF-Quantum energy filter by AutoEMation software, with the defocus from −1.2 to −1.5 μm and pixel size of 0.5141 Å. All of the micrographs were fractionated to 32 frames with a total irradiation dose of ~50 e-/Å2, which were then motion-corrected by MotionCor238 for following structural reconstruction. More details were summarized in Supplementary Table 1.The micrographs were then imported into CryoSPARC39 to perform patch CTF estimation and particle picking. Several rounds of 2D and 3D classification were applied to exclude bad particles. A final Non-uniform Refinement with local CTF refinement was used for high-resolution structure determination. Angular distribution plots were generated by RELION40. To build the atomic models, 6J6K, 7W9I, 7W92, and 3J9I from PDB databank were fetched to rigid-body fit into the maps of streptavidin, ACE2 (ACE2-RBD), Spike, and 20S proteasome in Chimera41, respectively, and then real-space refined in PHENIX42. All these cryo-EM data collection, refinement and validation statistics are summarized in Supplementary Table 1. Efficiency of angular distribution was calculated by cryo-EF28.To avoid the potential biases when comparing the orientation distribution from different datasets, we performed the reconstruction of experimental and control groups with identical procedures, parameters, and selection criteria in CryoSPARC39. During 3D reconstruction, the same reference was used. The final number of particles used for plotting Euler angle distribution was also identical.Cryo-ET data acquisition and analysisThe cryo-ET tilt series were acquired from +60 to −60° at the step of 3°, on a Thermo Fisher Scientific Titan Krios (300 kV) TEM equipped with a K3 camera (Gatan, Inc.) using SerialEM. The imaging defocus was set to −5 μm with a pixel size of 1.25 Å, and each tilt was fractionated to 8 frames with a total dose of ~3 e-/Å2. The beam-induced motion was corrected by MotionCor238, and then the tilt series were imported into IMOD43 for alignment and tomogram reconstruction, followed by denoising via IsoNet44. The positions of 20S proteasome were marked by template matching, as well as visual interpretation.Statistics and reproducibilityThe experiments in Figs. 2b, 2e, 2f, and 3g–i were repeated for more than 3 times independently with similar results, including the STM characterizations of GSAMs, SEM imaging of suspended GSAMs membranes, cryo-EM characterizations of suspended GSAMs, cryo-EM imaging of 20S proteasome particles with the presences of GSAMs, CHAPSO, and DDM.Reporting summaryFurther information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Self-assembled superstructure alleviates air-water interface effect in cryo-EM
