key: cord-296268-kb7fgfaq
authors: Mendonça, Luiza; Howe, Andrew; Gilchrist, James B.; Sun, Dapeng; Knight, Michael L.; Zanetti-Domingues, Laura C.; Bateman, Benji; Krebs, Anna-Sophia; Chen, Long; Radecke, Julika; Sheng, Yuewen; Li, Vivian D.; Ni, Tao; Kounatidis, Ilias; Koronfel, Mohamed A.; Szynkiewicz, Marta; Harkiolaki, Maria; Martin-Fernandez, Marisa L.; James, William; Zhang, Peijun
title: SARS-CoV-2 Assembly and Egress Pathway Revealed by Correlative Multi-modal Multi-scale Cryo-imaging
date: 2020-11-05
journal: bioRxiv
DOI: 10.1101/2020.11.05.370239
sha: 
doc_id: 296268
cord_uid: kb7fgfaq

Since the outbreak of the SARS-CoV-2 pandemic, there have been intense structural studies on purified recombinant viral components and inactivated viruses. However, investigation of the SARS-CoV-2 infection in the native cellular context is scarce, and there is a lack of comprehensive knowledge on SARS-CoV-2 replicative cycle. Understanding the genome replication, assembly and egress of SARS-CoV-2, a multistage process that involves different cellular compartments and the activity of many viral and cellular proteins, is critically important as it bears the means of medical intervention to stop infection. Here, we investigated SARS-CoV-2 replication in Vero cells under the near-native frozen-hydrated condition using a unique correlative multi-modal, multi-scale cryo-imaging approach combining soft X-ray cryo-tomography and serial cryoFIB/SEM volume imaging of the entire SARS-CoV-2 infected cell with cryo-electron tomography (cryoET) of cellular lamellae and cell periphery, as well as structure determination of viral components by subtomogram averaging. Our results reveal at the whole cell level profound cytopathic effects of SARS-CoV-2 infection, exemplified by a large amount of heterogeneous vesicles in the cytoplasm for RNA synthesis and virus assembly, formation of membrane tunnels through which viruses exit, and drastic cytoplasm invasion into nucleus. Furthermore, cryoET of cell lamellae reveals how viral RNAs are transported from double-membrane vesicles where they are synthesized to viral assembly sites; how viral spikes and RNPs assist in virus assembly and budding; and how fully assembled virus particles exit the cell, thus stablishing a model of SARS-CoV-2 genome replication, virus assembly and egress pathways.

Since the outbreak of the SARS-CoV-2 pandemic, there have been intense structural studies 24 on purified recombinant viral components and inactivated viruses. However, investigation of 25 the SARS-CoV-2 infection in the native cellular context is scarce, and there is a lack of 26 comprehensive knowledge on SARS-CoV-2 replicative cycle. Understanding the genome 27 replication, assembly and egress of SARS-CoV-2, a multistage process that involves different 28 cellular compartments and the activity of many viral and cellular proteins, is critically Krios to identify each individual infected cell (39.2 % for MOI of 0.1 and 45.4% for MOI 124 0.5) where cryoET tilt series were collected at the cell periphery. The grids were then 125 transferred to a FIB/SEM dualbeam instrument and the same infected cells were subjected to 126 either serial cryoFIB/SEM volume imaging (Zhu et al., 2021) or cryoFIB milling of cellular 127 lamellae where additional cryoET tilt series were collected (Sutton et al., 2020) . 128

Alternatively, we imaged infected cells on cryoEM grids by soft X-ray cryo-tomography 129 Figure 1D , Movie 5). We noticed that in a recent study, such 160 cytoplasm invagination was also seen in one of the conventional EM images of stained 161 plastic sections of SARS-CoV-2 infected cells, although no description was given (Lamers et 162 al., 2020) . 163 164 Independently, we investigated cytopathic effect of SARS-CoV-2 infection using soft X-ray 165 cryo-tomography. Consistent with the cryoFIB/SEM volume imaging results, the overview 166 images of soft X-ray display profound changes in mitochondria morphology, as they appear 167 fragmented in the SARS-CoV-2 infected cell ( Figure S3B&D on the cell surface are clearly distinguishable in soft X-ray cryo-tomogram ( Figure S3E , 174 black arrow). Also clearly observed are extensive vesiculation ( Figure S3F ) and a partial 175 nucleus invagination in the infected cell ( Figure 3G) . CoV-2 (total 9 portals from 24 DMVs) compared to those of MHV (8 portals per DMV), 207 signifying the difference between coronaviruses. We also observed vaultosomes near the 208 DMV outer membranes ( Figure 2F , black arrow). It is still unclear what is the physiological 209 role of vaults is, but they have been associated with RNA nucleocytoplasmic transport, innate suggests that proteolytic processing has taken place in these compartments resulting in S1 262

shedding. Therefore, we suggest that assembly at the ERGIC SMVs is the only pathway 263 which will lead to infectious viral progeny. 264

There has not been much detailed studies on how SARS-CoV-2 viruses are released from 267 cell. We investigated SARS-CoV-2 egress using both serial cryoFIB/SEM volume imaging 268 and cryoET. CryoFIB/SEM images clearly reveal virus exiting tunnels in 3D at the cell 269 periphery connecting to cell membrane ( Figure 5A -B, Movie 2). This likely resulted from the 270 fusion of very large multi-virus containing vesicles with cell membrane, i.e. egress through exocytosis. Consistent with cryoFIB/SEM analysis, we also observed virus exiting tunnels in 272 cryo-tomograms ( Figure 5C ). The fact that these compartments often contained many viral 273 particles suggests that this is a snapshot of viral exit, rather than cellular entry. 274

However, in addition to exocytosis, we also frequently found plasma membrane 276 discontinuities next to viral particles outside the cell ( Figure 5E We used a multi-modal, multi-scale and correlative approach to investigate SARS-CoV-2 305 infection process in native cells, from the whole cell to subcellular and to the molecular level. 306

The integration of multi-scale imaging data, achieved through this advanced workflow 307 ( Figure S1 ), has led us to propose a pathway for SARS-CoV-2 replication, in particular virus 308 genome replication, assembly and egress. The replication of SARS-CoV-2 appears spatially 309 well-organized and highly efficient. From genome replication, to protein synthesis and 310 transport, to virus assembly and budding, these processes take place in close-knit cytoplasmic 311 compartments. As illustrated in Figure The cells were incubated at room temperature for 15 minutes after which a further 1.5 mL of 446 media was added to each well. The plate was then incubated at 37 °C for 24 hours following 447 which supernatants were discarded and cells washed with 2 mL of PBS. The cells were then 448 fixed by addition of 3 mL of 4% paraformaldehyde in PBS for 1 hour at room temperature. 449

After fixation, grids were plunge-frozen on a Leica Grid plunger 2. 1 ul of concentrated 10 450 nm gold fiducials was applied to the gold-side of the EM grid and blotted from the gold-side. 451

The grid was quickly immersed in liquid ethane after blotting. Frozen grids were stored in 452 liquid nitrogen until data collection. Milling of SARS-Cov-2 infected cells was carried out using a Scios DualBeam cryoFIB 477 (ThermoFisher Scientific) equipped with a PP3010T transfer system and stage (Quorum 478 Technologies). Grids were sputter coated within the PP3010T transfer chamber maintained at 479 -175 °C. After loading onto the Scios stage at -168 °C, the grids were inspected using the 480 SEM (operated at 5 kV and 13 pA) and cells, identified as infected from TEM, were found. 481

The grid surface was coated using the gas injection system 482 (Trimethyl(methylcyclopentadienyl)platinum(IV), ThermoFisher Scientific) for 3 s, yielding 483 a thickness of ~3 um. Milling was performed using the ion beam operated at 30 kV and 484 currents decreasing from 300 pA to 30 pA. At 30 pA lamella thickness was less than 300 nm. 485

During the final stage of milling, SEM inspection of the lamellae was conducted at 2 kV and 486 13 pA. 487 488

Samples were imaged on a Zeis Crossbeam 550XL fitted with a Quorum transfer station and 490 cryo-stage. They were mounted on a Quorum-compatible custom sample holder and coated 491 with platinum for 60 sec at 10 mA on the Quorum transfer stage, prior to loading on the cryo-492 stage. Stage temperature was set at -165°C, while the anticontaminator was held at -185°C. 493

Samples were imaged at 45° tilt after being coated again with Pt for 2x 30sec using the FIB-494 SEM's internal GIS system, with the Pt reservoir held at 25°C. Initial trapezoid trenches were 495 milled at 30kV 7nA over 15 μm to reach a final depth of 10 μm, with a polish step over a 496 rectangular box with a depth of 10 μm performed at 30kV 1.5 nA. Serial Sectioning and 497

Imaging acquisition was performed as follows: FIB milling was set up using the 30kV 700 498 pA probe, a z-slice step of 20 nm and a depth of 10 μm over the entire milling box; SEM 499

imaging was performed at a pixel depth of 3024x2304 pixels, which resulted in a pixel size of 500 6.5 nm, with the beam set at 2kV 35pA, dwell time 100 nsec and scan speed 1, averaging the 501 signal over 100 line scans as a noise-reduction strategy. 502

CryoET image processing 504

The frames in each tilt angle in a tilt series were processed to correct drift using MotionCor2 505 (Zheng et al., 2017) . For the intact cells dataset, all tilt series were aligned using the default 506 parameters in IMOD version 4.10.22 with the eTomo interface, using gold-fiducial markers 507 (Kremer et al., 1996) . For lamella dataset, The tilt series were aligned in the framework of 508

Appion-Protomo fiducial-less tilt-series alignment suite (Noble and Stagg, 2015) . After tilt series alignment, the tilt-series stacks together with the files describing the projection 510 transformation and fitted tilt angles were transferred to emClarity for the subsequent sub-511 tomogram averaging analysis (Himes and Zhang, 2018) . 512 513

All sub-tomogram averaging analysis steps were performed using emClarity, mostly 515 following previously published protocols described workflow (Himes and Zhang, 2018) . The 516 CTF estimation for each tilt was performed by using emClarity version 1.4.3 , and the 517 subvolumes were selected by using automatic template matching function within emClarity 518 using reference derived from EMDB-21452 (Walls et al., 2020) that was low-pass filtered to 519 30-Å resolution in emClarity. The template matching results were cleaned manually by 520 comparison of the binned tomograms overlaid with the emClarity-generated IMOD model 521

showing the x,y,z coordinates of each cross-correlation peak detected. After manually 522 template cleaning, A total of 450 subvolumes from the lamella dataset and a total of 7090 523 subvolumes from the extracellular viruses dataset were retained, deriving from 3 tilt-series 524 and 50 tilt-series respectively, for the following averaging and alignment steps in emClarity. 525

For the extracellular viruses dataset, the 3D iterative averaging and alignment procedures 526 were carried out gradually with binning of 4x, 3x, 2x , each with 2-3 iterations with 527 increasingly restrictive search angles and translational shifts. 3-fold symmetry was applied 528 during all the steps. Final converged average map was generated using bin2 tomograms with 529 pixel size of 3.26 Å/pixel and a box size of 123×123×123 voxels. Resolution indicated by 530 0.143 FSC cut-off was 8.7 Å. The same process was carried out for lamella dataset, except 531 for the final average map was generated with pixel size of 4.26 Å/pixel 532 and a box size of 90×90×90 voxels and a final resolution at 11 Å (Gold standard FSC at 533 0.143 cut-off). 534

Cell structures were manually segmented from stacks of images using ImageJ (Koppensteiner 537 et al., 2012) (Kounatidis et al., 2020) . Grids were loaded on the X-ray microscope at 553 B24 and were first mapped using visible light with a 20X objective. The resulting coordinate-554 map was used to locate areas of interest where 2D X-ray mosaics were collected (X-ray light 555 used was at 500eV) and used to identify areas of interest within. Tilt series of 100-120 º were 556 collected for each field of view area of interest at 0.2 or 0.5º steps with constant exposure of 557 0.5 sec keeping average pixel intensity to between 5-30k counts. All tilt series were background subtracted, saved as raw Tiff stacks and reconstructed using either IMOD 559 (Kremer et al., 1996) or Batchruntomo (Mastronarde, 2005) . 

The 

Membrane rearrangements mediated 614 by coronavirus nonstructural proteins 3 and 4. Virology 458-459

Cryo-soft X-ray tomography: using soft X-rays to explore the 617 ultrastructure of whole cells

emClarity: software for high-resolution 619 cryo-electron tomography and subtomogram averaging

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by 624 a Clinically Proven Protease Inhibitor

5'-Triphosphate 627 RNA is the ligand for RIG-I

Structures and 630 distributions of SARS-CoV-2 spike proteins on intact virions

SARS-633 coronavirus replication is supported by a reticulovesicular network of 634 modified endoplasmic reticulum

Macrophage internal HIV-1 is protected from 637 neutralizing antibodies

3D Correlative Cryo-Structured Illumination 641 Fluorescence and Soft X-ray Microscopy Elucidates Reovirus Intracellular 642

Computer 644 visualization of three-dimensional image data using IMOD

SARS-CoV-2 productively infects human gut enterocytes

Structure of the SARS-CoV-2 spike 652 receptor-binding domain bound to the ACE2 receptor

The SARS-CoV-2 Nucleocapsid phosphoprotein forms mutually exclusive condensates with RNA and the membrane-associated M protein. bioRxiv : the 656 preprint server for biology

Automated electron microscope tomography using 662 robust prediction of specimen movements

The Hepatitis C Virus-665 Induced Membranous Web and Associated Nuclear Transport Machinery Limit 666 Access of Pattern Recognition Receptors to Viral Replication Sites

A 670 structural analysis of M protein in coronavirus assembly and morphology

Automated batch fiducial-less tilt-673 series alignment in Appion using Protomo

SARS-coronavirus-2 replication in Vero E6 678 cells: replication kinetics, rapid adaptation and cytopathology

Expression and cleavage 682 of middle east respiratory syndrome coronavirus nsp3-4 polyprotein induce 683 the formation of double-membrane vesicles that mimic those associated with 684 coronaviral RNA replication

Morphological and Biochemical Characterization of the Membranous 687

Hepatitis C Virus Replication Compartment

UCSF Chimera--a visualization system 691 for exploratory research and analysis

ER-derived vesicles 695 exporting short-lived ERAD regulators, for replication

Structural basis of receptor recognition 699 by SARS-CoV-2

Cardiomyocytes Are Susceptible to SARS-CoV-2 Infection

Characterization of severe acute respiratory syndrome-706 associated coronavirus (SARS-CoV) spike glycoprotein-mediated viral entry

A unifying structural and functional model of the 712 coronavirus replication organelle: Tracking down RNA synthesis

Cryo-EM structure of the 715 SARS coronavirus spike glycoprotein in complex with its host cell receptor

The intracellular sites 719 of early replication and budding of SARS-coronavirus

Structural model of the SARS 722 coronavirus E channel in LMPG micelles

Assembly intermediates of 726 orthoreovirus captured in the cell

Free fatty acid binding pocket in the locked structure of SARS

CoV-2 spike protein. Science

In situ 733 structural analysis of SARS-CoV-2 spike reveals flexibility mediated by 734 three hinges

Nucleocapsid-independent 737 assembly of coronavirus-like particles by co-expression of viral envelope 738 protein genes

Structure, Function, and Antigenicity of the SARS-CoV

Structural and Functional Basis of 744 SARS-CoV-2 Entry by Using Human ACE2

A molecular pore spans the double membrane of the 748 coronavirus replication organelle

Membrane Vesicles as Platforms for Viral Replication. Trends in 751 Microbiology

Direct 753 visualization of vaults within intact cells by electron cryo-tomography

Structural basis for the recognition of the SARS-CoV-2 by full-length human 757 ACE2

Molecular architecture of the SARS-CoV-2 760 virus

MotionCor2: anisotropic correction of beam-induced 763 motion for improved cryo-electron microscopy

Short 770 Article Serial cryoFIB / SEM Reveals Cytoarchitectural Disruptions in Leigh 771 Syndrome Patient Cells Serial cryoFIB / SEM Reveals Cytoarchitectural 772 Disruptions in Leigh Syndrome Patient Cells