key: cord-103739-mmkrwj8t
authors: Snijder, Eric J.; Limpens, Ronald W.A.L.; de Wilde, Adriaan H.; de Jong, Anja W. M.; Zevenhoven-Dobbe, Jessika C.; Maier, Helena J.; Faas, Frank F.G.A.; Koster, Abraham J.; Bárcena, Montserrat
title: A unifying structural and functional model of the coronavirus replication organelle: tracking down RNA synthesis
date: 2020-03-24
journal: bioRxiv
DOI: 10.1101/2020.03.24.005298
sha: 
doc_id: 103739
cord_uid: mmkrwj8t

Zoonotic coronavirus (CoV) infections, like those responsible for the current SARS-CoV-2 epidemic, cause grave international public health concern. In infected cells, the CoV RNA-synthesizing machinery associates with modified endoplasmic reticulum membranes that are transformed into the viral replication organelle (RO). While double-membrane vesicles (DMVs) appear to be a pan-coronavirus RO element, studies to date describe an assortment of additional coronavirus-induced membrane structures. Despite much speculation, it remains unclear which RO element(s) accommodate viral RNA synthesis. Here we provide detailed 2D and 3D analyses of CoV ROs and show that diverse CoVs essentially induce the same membrane modifications, including the small open double-membrane spherules (DMSs) previously thought to be restricted to gamma- and delta-CoV infections and proposed as sites of replication. Metabolic labelling of newly-synthesized viral RNA followed by quantitative EM autoradiography revealed abundant viral RNA synthesis associated with DMVs in cells infected with the beta-CoVs MERS-CoV and SARS-CoV, and the gamma-CoV infectious bronchitis virus. RNA synthesis could not be linked to DMSs or any other cellular or virus-induced structure. Our results provide a unifying model of the CoV RO and clearly establish DMVs as the central hub for viral RNA synthesis and a potential drug target in coronavirus infection.

epidemic, cause grave international public health concern. In infected cells, the CoV RNA-23 synthesizing machinery associates with modified endoplasmic reticulum membranes that are 24 transformed into the viral replication organelle (RO). While double-membrane vesicles 25 (DMVs) appear to be a pan-coronavirus RO element, studies to date describe an assortment RO [15, 20, 21] , was entirely possible and started to attract 71 attention. Notably, DMVs can be also formed in the absence of vRNA synthesis by 72 expression of key transmembrane nsps [22] [23] [24] [25] [26] . Moreover, several studies suggested a lack of 73 direct correlation between the number of DMVs and the level of CoV replication in the 74 infected cell [27, 28] . 75 The interpretation of the CoV RO structure and function was further compounded by the 76 discovery of different RO 

106 double-membrane spherules (DMSs) 107 We first set out to analyse the ultrastructure of MERS-CoV-infected Huh7 cells under sample 108 preparation conditions favourable for autoradiography (see Materials and Methods) (Fig 1, S1 109 Video). Strikingly, in addition to the DMVs and CM that are well established hallmarks of 110 beta-CoV infections, the presence of small spherules, occasionally in large numbers, was 111 readily apparent (Fig 1A and 1B) . These spherules were notably similar to the DMSs 112 previously described for the gamma-CoV IBV [29] . Their remarkably regular size of ~80 nm 113 (average diameter 79.8 ± 2.5 nm, n=58), a delimiting double membrane and their electron-114 dense content, made these spherules clearly distinct from other structures, including progeny 115 virions, which had comparable diameter (Fig 1C and 1D ). The double-membrane spherules 116 (DMSs) generated during IBV infection were previously described as invaginations of the 117 zippered ER that remain open to the cytosol [29] . In MERS-CoV-infected cells, the DMSs 118 were connected to the CM from which they seemed to derive ( Fig 1E) . Clear openings to the 119 cytosol could not be detected for the large majority (~80%, n=54) of the fully reconstructed 120 DMSs, which suggests that the original invagination may eventually transform into a sealed 121 compartment. This type of apparently closed DMSs were also present, though in a lower 122 proportion (~50%, n=39), in IBV-infected cell samples processed in an identical manner (S1 The 3D architecture of MERS-CoV-induced RO aligned with previous observations for other 156 CoV [15, 29] . No clear openings connecting the interior of the DMVs and the cytosol could 157 be detected. All three types of MERS-CoV-induced membrane modifications appeared to be 158 interconnected, either directly or indirectly through the ER. While DMSs were connected to 159 CM, and CM to ER, ER membranes were often continuous with DMVs (Fig 1F, arrowheads) . 160 Therefore, like other CoVs, MERS-CoV infection appears to induce a network of largely 161 interconnected modified ER membranes that, as a whole, can be considered the CoV RO. Quantification of the autoradiography signal per subcellular structure (see also S1 Table) . Labelling 214 densities and relative labelling indexes (RLI) in different subcellular regions of (C) Vero E6 cells 215 infected with SARS-CoV (MOI 10) or (D) Huh7 cells infected with MERS-CoV (MOI 5). 216

Radioactively-labelled uridine was provided for the indicated periods of time immediately before 217 fixation at 7 hpi and 12 hpi, respectively. Control mock-infected cells are excluded from the RLI 218 plots, as RLI comparisons between conditions require the same number of classes (subcellular 219 regions) and these cells lack ROs and virions. budding vesicles (Fig 6A-C) .The M and S proteins also localized to the Golgi complex, 348 aligning with previous observations for other CoVs [20, 44, 48, 49] . The MERS-CoV N 349 protein was found in regions with CM and DMSs, though the distribution of signal was 350 homogenous and DMSs were not particularly densely labelled (Fig 6D) . The presence of the assembly, or the S protein (Fig 6E and 6F ). 356 Previously, the CM induced by SARS-CoV and MHV were shown by IEM to accumulate 357 viral nsps, while dsRNA signal was primarily found inside the DMVs [15, 20] . Similarly, 358 nsp3 mapped to the CM induced in MERS-CoV infection, but also to the DMSs to a 359 comparable extent (Fig 6G) . Our attempts to combine dsRNA antibody labelling with thawed 360 cryo-sections were unsuccessful, which made us resort to HPF-FS samples. In these, 361 however, while DMVs were easily detected, the morphology of CM and DMSs was less 362 clearly defined. Nevertheless, dsRNA signal was clearly associated with DMVs, while the 363 dark membranous regions between DMVs that we interpreted as CM and DMSs clusters 364 appeared devoid of signal (Fig 6H and 6I ). 365 In summary, for the antibodies tested (recognizing N, M, S, nsp3, and dsRNA), the labelling 366 pattern in MERS-CoV-induced DMSs closely resembled that of the CM, from which they 367 seem to derive. The absence of labelling for key proteins in virus assembly, like the M and S 368 proteins, strongly suggest that DMSs do not represent (spurious) virus assembly events.

The comprehensive analysis presented here demonstrates that viruses across different CoV 371 genera induce essentially the same type of membrane structures. After somewhat disparate 372 observations [15, 20, 21, 29, 30, 47] , the unifying model that emerges from our study is that Our results add to studies that, in the last years and after much speculation, have started to 419 provide experimental evidence that the DMVs induced by +RNA viruses are active sites of 420 vRNA synthesis [11, [58] [59] [60] . However, it is not clear that DMVs always play the primary role 421 in virus replication that we demonstrate here for CoV. For picornaviruses, for example, virus- were freeze-substituted in a Leica AFS2 system with 0.1% (wt/vol) uranyl acetate as 576 previously described [71] , with the only modification that acetone was replaced by ethanol 577 from the last washing step before Lowycril infiltration onwards. Cell sections (75 nm thick) 578 were incubated with the primary mouse antibody, then with a bridging rabbit anti-mouse-IgG 579 antibody (Dako Cytomation), and finally with protein A coupled to 15-nm gold particles. 580 After immunolabelling, samples were additionally stained with 7% uranyl acetate and 581 Reynold's lead citrate. Large mosaic EM maps containing dozens of cell profiles were used for the quantitative 608 analysis of the newly-synthesized RNA autoradiography signal (see S1 Table) . For each 609 CoV, different conditions (infected and mock-infected cells, plus different labelling times) 610 were compared using only samples developed after the same period of time. The analysis of 611 the signal in different subcellular regions was carried out using home-built software. Areas of 612 4 µm 2 were randomly selected from the mosaic EM maps and the autoradiography grains 613 present in those areas were manually assigned to the underlying cellular structures. The 614 abundance of the different types of subcellular structures was estimated through virtual points 615 in a 5×5 lattice superimposed to each selected area, which were also assigned to the different 616 subcellular classes. Regularly along the process, the annotated data per condition was split 617 into two random groups and the Kendall and Spearman coefficients, which measure the 618 concordance between two data sets [73], were calculated. New random regions were added 619 until the average Kendall and Spearman coefficients resulting from 10 random splits were 620 higher than 0.8 and 0.9, respectively (maximum value, 1). Labelling densities and relative 621 labelling indexes (RLI) were then calculated from the annotated points [39] . 622 For the analysis of the association of vRNA synthesis with each of the different ROs motifs, 623 the specific DMVs, DMSs and CM included in the analysis were carefully selected. Only 624 individual DMVs that were at least one micron away from any other virus-induced membrane 625 modification were included in the analysis. For every grain present in an area of 750 nm 626 radius around each DMV, the distance to the DMV centre was measured. In the case of 627 DMSs, which were always part of clusters of virus-induced membrane structures, only DMSs 628 in the periphery of these clusters were selected. The quantified signal was limited to sub-629 areas devoid of other RO motifs, which were defined by circular arcs (typically 30 o to 100 o , 630 radius 500 nm) opposite to the RO clusters. CMs are irregular structures that appear partially 631 28 or totally surrounded by DMVs. Only large CM (> 0.6 µm across) were selected in order to 632 make more apparent (if present) any decay of the autoradiography signal as the distance to 633 the surrounding DMVs increased. For each autoradiography grain, both the distance to the 634 closest CM boundary (d 1 ) and the distance to the opposite CM edge (d 2 ) were measured. The 635 relative distance to the CM edge was then calculated as d 1 /(d 1 +d 2 ) and expressed in 636 percentages. All the measurements in different DMVs, DMSs and CM were made using 637 Aperio Imagescope software (Leica) and pooled together into three single data sets. Autoradiography is a classic technique that allows the EM visualization of a radioactive 3 marker, usually targeting a certain process, and thus reveals the subcellular localization of 4 that process [1, 2] . Tritiated uridine, for example, can be used to locate active RNA synthesis 5 [3] [4] [5] , as shown also in this study. A clear advantage over the use of alternatives for metabolic 6 labelling of newly-synthesized RNA (e.g. Br-uridine, Br-UTP, 5-ethynil uridine) is that the 7 radioactive precursor is chemically identical to the natural substrate. 8 After labelling, the samples are immediately fixed and processed for EM. The location of the 9 radioactive marker can then be made apparent by applying a highly-sensitive photographic 10 emulsion (a nuclear emulsion) on top of the cell sections and exposing it for several weeks to 11 months. The beta particles that are emitted as a result of tritium disintegrations generate 12 electrons that get trapped in the silver halide emulsion and create a "latent image". When the 13 emulsion is developed, these negative charges promote the reduction to metallic silver, 14 generating electron-dense grains that are visible by EM. In principle, given enough time to 15 accumulate enough radioactive disintegrations, even low levels of the radioactive marker 16 could be detected. In practice, other factors (e.g. background radiation, emulsion aging) set 17 some limits to autoradiography, which is nonetheless a very sensitive technique. 18 The resolution of EM autoradiography is limited by the fact that radioactive disintegrations 19 generate beta particles that are emitted in random directions. Importantly, the probability of 20 giving rise to signal degreases with the distance from the radioactive source; however, some 21 beta particles may travel up to a few hundred nanometers before striking the photographic 22 emulsion [2] . Therefore, it is important to keep in mind that the silver grains may not directly 23 overlay the structure containing the radioactive source. Quantitative analyses of the signal 24 2 that take this factor into account, like those presented in this study, become indispensable to 25 maximize the information that autoradiography can provide. as for MER-CoV-infected cells (Fig 1) . Tomographic slices through two regions containing 32

Membranous replication factories induced by plus-646 strand RNA viruses

Ultrastructure of the replication sites of positive-strand RNA 649 viruses

Building Viral Replication Organelles

Close Encounters of the Membrane Types

Interaction of the innate immune system with positive-656 strand RNA virus replication organelles

RNA virus replication complex parallels form and function of retrovirus capsids. Molecular 660 cell

Three-dimensional 662 analysis of a viral RNA replication complex reveals a virus-induced mini-organelle. PLoS 663 biology

Composition and three-dimensional architecture of the dengue virus replication and assembly 667 sites

Template RNA 670 length determines the size of replication complex spherules for Semliki Forest virus

Three-674 dimensional imaging of the intracellular assembly of a functional viral RNA replicase 675 complex

The transformation of enterovirus replication structures: a three-dimensional study 679 of single-and double-membrane compartments

Complex dynamic 682 development of poliovirus membranous replication complexes

Membrane alterations induced by nonstructural proteins of human norovirus

Three-690 dimensional architecture and biogenesis of membrane structures associated with hepatitis C 691 virus replication

RNA replication of 694 mouse hepatitis virus takes place at double-membrane vesicles

SARS-coronavirus replication is supported by a reticulovesicular network of 698 modified endoplasmic reticulum

Ultrastructural characterization of arterivirus replication structures: reshaping the 703 endoplasmic reticulum to accommodate viral RNA synthesis

Extensive 707 coronavirus-induced membrane rearrangements are not a determinant of pathogenicity

An integrated analysis of 711 membrane remodeling during porcine reproductive and respiratory syndrome virus 712 replication and assembly

Double-stranded

RNA is produced by positive-strand RNA viruses and DNA viruses but not in detectable 717 amounts by negative-strand RNA viruses

Qualitative and quantitative 721 ultrastructural analysis of the membrane rearrangements induced by coronavirus. Cellular 722 microbiology

MERS-coronavirus replication induces severe in vitro cytopathology and is 726 strongly inhibited by cyclosporin A or interferon-alpha treatment

Severe acute 730 respiratory syndrome coronavirus nonstructural proteins 3, 4, and 6 induce double-membrane 731 vesicles

Mutations across murine 738 hepatitis virus nsp4 alter virus fitness and membrane modifications

Expression 742 and Cleavage of Middle East Respiratory Syndrome Coronavirus nsp3-4 Polyprotein Induce

the Formation of Double-Membrane Vesicles That Mimic Those Associated with Coronaviral 744 RNA Replication

Bronchitis Virus Nonstructural Protein 4 Alone Induces Membrane Pairing

Competitive 751 fitness in coronaviruses is not correlated with size or number of double-membrane vesicles 752 under reduced-temperature growth conditions

Targeting membrane-bound viral RNA synthesis reveals potent inhibition of diverse 757 coronaviruses including the middle East respiratory syndrome virus

Infectious bronchitis virus generates spherules from zippered endoplasmic reticulum 762 membranes

Replication Organelle Comprises Double-Membrane Vesicles and Zippered Endoplasmic 766

Does form meet function in the 769 coronavirus replicative organelle? Trends in microbiology

Isolation of 772 a novel coronavirus from a man with pneumonia in Saudi Arabia. The New England journal 773 of medicine

Genomic characterization of a newly discovered coronavirus associated with acute 777 respiratory distress syndrome in humans

The New England journal of medicine

Genomic characterisation and 783 epidemiology of 2019 novel coronavirus: implications for virus origins and receptor binding

Techniques and applications of autoradiography in the light and electron 787 microscope

Electron Microscopy: 789 Principles and Techniques for Biologists

Virtual nanoscopy: generation of ultra-large high resolution electron microscopy maps

Relative labelling index: a novel stereological 796 approach to test for non-random immunogold labelling of organelles and membranes on 797 transmission electron microscopy thin sections

Replication of coronavirus MHV-A59 in sac-cells: 800 determination of the first site of budding of progeny virions

Ultrastructural characterization of SARS coronavirus

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

Novel 811 contrasting and labeling procedures for correlative microscopy of thawed cryosections

Coronavirus M proteins accumulate in the Golgi complex beyond the site of virion budding

Proliferative growth of SARS 820 coronavirus in Vero E6 cells

Morphogenesis of avian infectious 823 bronchitis virus and a related human virus (strain 229E)

Ultrastructural 826 Characterization of Membrane Rearrangements Induced by Porcine Epidemic Diarrhea Virus 827

The E1 glycoprotein of an 830 avian coronavirus is targeted to the cis Golgi complex

The 838 putative helicase of the coronavirus mouse hepatitis virus is processed from the replicase 839

gene polyprotein and localizes in complexes that are active in viral RNA synthesis

Localization of mouse hepatitis virus nonstructural proteins and RNA synthesis indicates a 844 role for late endosomes in viral replication

Mouse hepatitis virus replicase proteins 847 associate with two distinct populations of intracellular membranes

Determination of host proteins composing the microenvironment of coronavirus replicase 852 complexes by proximity-labeling. eLife

The coronavirus nucleocapsid is a 855 multifunctional protein

Visualizing 859 coronavirus RNA synthesis in time by using click chemistry

Do viruses subvert cholesterol 863 homeostasis to induce host cubic membranes? Trends in cell biology

Cubic membranes: a legend beyond the

Flatland* of cell membrane organization

Morphological and 870 biochemical characterization of the membranous hepatitis C virus replication compartment

Epub 2013/07/26

Escaping Host Factor PI4KB Inhibition: Enterovirus Genomic RNA Replication in the 875 Absence of Replication Organelles

878 The Origin, Dynamic Morphology, and PI4P-Independent Formation of 879

Encephalomyocarditis Virus Replication Organelles. mBio

SARS-coronavirus replication/transcription complexes are membrane-protected 884 and need a host factor for activity in vitro

Early 888 endonuclease-mediated evasion of RNA sensing ensures efficient coronavirus replication

Mechanisms and 892 enzymes involved in SARS coronavirus genome expression

A new virus isolated from the human respiratory tract

Reverse genetics system for 898 the avian coronavirus infectious bronchitis virus

The coronavirus spike protein is a 902 class I virus fusion protein: structural and functional characterization of the fusion core 903 complex

Ultrastructure and origin of membrane vesicles associated with the severe 907 acute respiratory syndrome coronavirus replication complex

Monoclonal 911 antibodies to double-stranded RNA as probes of RNA structure in crude nucleic acid extracts

Towards a solution to MERS: protective human monoclonal antibodies targeting 916 different domains and functions of the MERS-coronavirus spike glycoprotein

Resolution of a gold latensification-elon 920 ascorbic acid developer for Ilford L4 emulsion

Correlated 923 fluorescence and 3D electron microscopy with high sensitivity and spatial precision

A rapid 930 method for assessing the distribution of gold labeling on thin sections

DMSs (white 33 arrowheads) and zippered ER (white arrows). Most zippered ER consists of long stretches of 34

though branching zippered ER, closer to the CM 35 described for beta-CoV, was also present (B) Virus particles (black arrowheads) budding into 36 the ER membranes were often observed. Scale bars

Analysis 39 of previously described samples of CoV-infected cells, prepared for EM either by HPF(A) or 40 cryo-plunging (B). A targeted search revealed the presence of DMSs (white arrowheads) in 41 close association with CM. In comparison with the chemically fixed samples used in this 42 study, the superior ultrastructural preservation of cryo-fixation results in less distorted 43 membranes, but also in a denser cytoplasm and darker CM that makes DMS less apparent. 44 (A) Example from a MERS-CoV-infected Huh7 cell

SARS-CoV-infected cell (8 hpi)

Metabolic labelling of newly-synthesized vRNA in IBV-infected cells and 48 analysis of the autoradiography signal. Vero cells infected with IBV were pre-treated with 49

actinomycin D for 1 hour, then labelled for 30 or 60 min with tritiated uridine

A) Overview of 51 an IBV-infected Vero cell (60 min labelling). The areas containing DMVs and zippered ER 52 are outlined in yellow and blue, respectively, and other subcellular structures annotated (N, 53 nucleus; m, mitochondria; Au, autophagosome; VCR, virion-containing regions). The 54 autoradiography signal accumulates in areas of virus-induced membrane modifications that 55 often only contain DMVs

Close-up of the area boxed in black in (A), which contains DMVs

The contrast between the densely labelled DMVs and the 58 zippered ER and DMSs largely lacking signal is apparent and suggests that the 59 autoradiography grains sometimes present on the latter structures arose from radioactive 60 disintegrations in the surrounding active DMVs. (C) In agreement with this possibility, most 61 of the DMSs (96%) were devoid of signal, and most of those that contained label where close 62 to an active DMV (n DMS = 178). (D) Furthermore, the distribution of autoradiography grains 63 around DMSs resembled that of a random distribution

DMVs proved that these structures are associated with vRNA synthesis, as the signal reaches 66 maximum values in the proximity of the DMVs (n DMVs = 51). ((C, D) See Materials and 67 Methods for the selection criteria and details)

Electron tomography of the membrane structures induced in MERS-CoV infection

Animation illustrating the tomography reconstruction and model presented in Fig 1B. The 72 video first shows the tomographic slices (1.2 nm thick) through the reconstructed volume, 73 and then surface-rendered models of the different structures segmented from the tomogram

CM (blue) and DMVs (yellow and lilac, outer and inner membranes), ER 75 4 (green), and a vesicle (silver) containing virions (pink). The movie highlights the DMS 76 association with CM

Techniques and applications of autoradiography in the light and electron 84 microscope

Autoradiography & Radioautography. Electron Microscopy: 86 Principles and Techniques for Biologists

Association of polioviral proteins of the P2 89 genomic region with the viral replication complex and virus-induced membrane synthesis as 90 visualized by electron microscopic immunocytochemistry and autoradiography

Escaping Host Factor PI4KB Inhibition: Enterovirus Genomic RNA Replication in the 94 Absence of Replication Organelles

97 The Origin, Dynamic Morphology, and PI4P-Independent Formation of 98

Encephalomyocarditis Virus Replication Organelles. mBio

MERS-coronavirus replication induces severe in vitro cytopathology and is 103 strongly inhibited by cyclosporin A or interferon-alpha treatment

SARS-coronavirus replication is supported by a reticulovesicular network of 108 modified endoplasmic reticulum

 Table   79 Data sets and sampling for the quantifications of the autoradiography signal presented 80 in Fig 3C and