key: cord-0787966-msngrxsl
authors: Schmidt, Nora; Lareau, Caleb A.; Keshishian, Hasmik; Melanson, Randy; Zimmer, Matthias; Kirschner, Luisa; Ade, Jens; Werner, Simone; Caliskan, Neva; Lander, Eric S.; Vogel, Jörg; Carr, Steven A.; Bodem, Jochen; Munschauer, Mathias
title: A direct RNA-protein interaction atlas of the SARS-CoV-2 RNA in infected human cells
date: 2020-07-15
journal: bioRxiv
DOI: 10.1101/2020.07.15.204404
sha: 5c27876dd85f551d53138d9af31a208c763f7b0a
doc_id: 787966
cord_uid: msngrxsl

SARS-CoV-2 infections pose a global threat to human health and an unprecedented research challenge. Among the most urgent tasks is obtaining a detailed understanding of the molecular interactions that facilitate viral replication or contribute to host defense mechanisms in infected cells. While SARS-CoV-2 co-opts cellular factors for viral translation and genome replication, a comprehensive map of the host cell proteome in direct contact with viral RNA has not been elucidated. Here, we use RNA antisense purification and mass spectrometry (RAP-MS) to obtain an unbiased and quantitative picture of the human proteome that directly binds the SARS-CoV-2 RNA in infected human cells. We discover known host factors required for coronavirus replication, regulators of RNA metabolism and host defense pathways, along with dozens of potential drug targets among direct SARS-CoV-2 binders. We further integrate the SARS-CoV-2 RNA interactome with proteome dynamics induced by viral infection, linking interactome proteins to the emerging biology of SARS-CoV-2 infections. Validating RAP-MS, we show that CNBP, a regulator of proinflammatory cytokines, directly engages the SARS-CoV-2 RNA. Supporting the functional relevance of identified interactors, we show that the interferon-induced protein RYDEN suppresses SARS-CoV-2 ribosomal frameshifting and demonstrate that inhibition of SARS-CoV-2-bound proteins is sufficient to manipulate viral replication. The SARS-CoV-2 RNA interactome provides an unprecedented molecular perspective on SARS-CoV-2 infections and enables the systematic dissection of host dependency factors and host defense strategies, a crucial prerequisite for designing novel therapeutic strategies.

At the end of 2019, the rapid spread of a novel severe acute respiratory syndrome-related coronavirus (SARS-CoV-2) around the globe has led to a worldwide spike in a SARS-like respiratory illness termed Coronavirus Disease 2019 (COVID- 19) 1 . Due to the absence of effective antiviral therapy, COVID-19 has taken hundreds of thousands of lives to date and resulted in unprecedented socioeconomic disruptions.

A prerequisite for understanding SARS-CoV-2 infections and enabling novel therapeutic strategies is obtaining a detailed map of the molecular events and perturbations occurring as SARS-CoV-2 infects human host cells. SARS-CoV-2 is an enveloped, positivesense, single-stranded RNA virus that, upon infection of a host cell, deploys a 'translationready' RNA molecule, which engages the protein synthesis machinery of the host in order to express a limited number of viral proteins critical for its replication 2 . Thus, similar to other RNA viruses, SARS-CoV-2 is inherently dependent on recruiting host cell factors and machinery, including regulators of RNA stability, localization, and translation, to facilitate virus replication and the production of viral progeny. For the host cell, on the other hand, it is crucial to detect the presence of a viral pathogen and activate appropriate innate immune response pathways 3, 4 . To understand this interplay between virus and host, it is essential to characterize with molecular detail which host proteins make direct contact with viral RNA and may function as host dependency factors or antiviral regulators. To date, studies on SARS-CoV-2 infected human cells have primarily focused on characterizing expression or modification changes in the host cell transcriptome [5] [6] [7] or proteome [8] [9] [10] . While several studies described protein-protein interactions of recombinantly expressed viral proteins in uninfected cells 10, 11 , no study has comprehensively identified direct interactions between viral RNA and the host cell proteome in infected human cells.

To improve our understanding of the host factors that contribute to the regulation of SARS-CoV-2 RNA during infection, we sought to obtain an unbiased picture of the cellular proteins that directly bind to the SARS-CoV-2 RNA in infected human cells. Recent advances in RNA capture and quantitative mass spectrometry approaches [12] [13] [14] have made this endeavor highly tractable. Among the available technologies, we focused on those that use ultraviolet (UV) crosslinking to create covalent bonds between RNA and directly bound proteins, as opposed to chemical crosslinkers that also stabilize indirect interactions 15 .

Further, we wanted to ensure that the identified proteins bind directly to the SARS-CoV-2 RNA, excluding approaches that assess differential interactions across all cellular RNAs in response to infection. To satisfy these requirements, we selected RNA antisense purification and quantitative mass spectrometry (RAP-MS) 12, 13 , which implements a denaturing purification procedure to capture and identify proteins that crosslink directly to the SARS-CoV-2 RNA.

Using RAP-MS, we globally identify host cell proteins that directly bind to the SARS-CoV-2 RNA in infected human cells and integrate these binding events with proteome abundance changes induced by viral infection. Our work highlights the molecular interactions that underlie both virus replication and host defense mechanisms and further establishes that therapeutic inhibition of direct RNA binders can modulate SARS-CoV-2 replication.

To purify the SARS-CoV-2 RNA and the complement of directly crosslinked cellular proteins from virally infected human cells, we designed a pool of biotinylated DNA oligonucleotides antisense to the positive-sense SARS-CoV-2 RNA, such that one probe binding site occurs roughly every 400 bases in the ∼30 kb SARS-CoV-2 genome. As a cellular system, we selected the human HuH-7 cell line. In addition to being permissive to both SARS-CoV-1 and SARS-CoV-2 replication 16, 17 , the cellular proteins bound to all polyadenylated RNAs (poly(A)-RNA) have been identified in these cells 18 . Further, a previous study employed chemical crosslinking and RNA antisense purification in a HuH-7-derived cell line to identify proteins that directly or indirectly associate with the genomic RNAs of Dengue and Zika viruses 19 , two unrelated positive-sense RNA viruses.

To test if our pool of antisense capture probes was suitable for the purification of the SARS-CoV-2 RNA from infected HuH-7 cells, we performed RAP-MS 24 hours after infection.

Expanding upon the established RAP-MS procedure 13 , we implemented a covalent protein capture step following the release of SARS-CoV-2-bound proteins, which allowed us to identify the RNA sequences crosslinked to purified proteins ( Figure 1a , see Methods). While this approach yielded near-complete sequence coverage of the SARS-CoV-2 genome in RNA antisense purifications, most of the signal was observed near probe binding sites (Supplementary Figure 1a) . Importantly, sequencing reads originating from SARS-CoV-2 RNA made up 93% and 92% of all mapped reads in two replicate experiments (Supplementary Figure 1b) .

To identify proteins that specifically interact with SARS-CoV-2 RNA, we compared the protein content of SARS-CoV-2 purifications to that of an unrelated control ribonucleoprotein complex of known composition. As the control, we selected the human RNA Component of Figure 1d) . Together, these experiments demonstrate the specificity of the RAP approach for capturing the desired RNAs and their known direct protein binding partners.

Encouraged by the above results, we subjected proteins purified with both RMRP and SARS-CoV-2 RNAs to tandem mass tag (TMT) labeling and relative quantification by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). In two biological replicate experiments, we identified 699 proteins, of which 583 were detected with two or more unique peptides (Supplementary Table 1 ). As shown in Figure 1b , we found five known RMRP components among the 10 most significantly enriched proteins in RMRP purifications (Supplementary Table 1 ). This is consistent with our previous results 13 As expected, the SARS-CoV-2 N protein, which is thought to occupy the majority of the viral RNA, was one of the two most strongly enriched viral proteins, followed by several known viral RNA-binders, such as the endoribonuclease NSP15 22 , the RNA-dependent RNA polymerase (RdRP) NSP12 23 , the methyltransferase NSP16 24 , the single-stranded RNAbinding protein NSP9 25 , the capping factor NSP10 24 , the primase NSP8 23 , and the multifunctional protein NSP3 26 . In addition to NSPs, we also found ORF3A, which binds the 5'-end of the SARS-CoV-1 genomic RNA 27 , as well as the Spike protein (S), and the Membrane protein (M) among strongly enriched proteins.

While the M protein is known to interact with the RNA-binding protein N, a model for packaging of the coronavirus genomic RNA further suggests a possible direct RNA-binding function for M 28 . An RNA-binding activity of the S protein was not previously reported. While S covers the surface of the viral envelope, it has a transmembrane domain and an intracellular tail, 29 making it conceivable that S may indeed contact viral RNA. Similarly, NSP3 and NSP6 are thought to be necessary for the formation of cytoplasmic double-membrane vesicles 30 , which contribute to viral replication or virion assembly and may contact RNA as part of these functions.

Finally, we also detected a protein corresponding to a newly annotated out-of-frame internal ORF (iORF) within the ORF of the N protein (N.iORF1) 31 . N.iORF1 shares 72% of its amino acids with ORF9b of SARS-CoV-1, making it likely that both proteins are homologous.

N.iORF1 was moderately enriched in SARS-CoV-2 purifications (Figure 1b, c) , suggesting that it may interact with viral RNA, which raises the possibility that at least two different RNAbinding proteins are translated from overlapping ORFs within the sequence of the viral N gene.

We next focused on the human proteins enriched in SARS-CoV-2 purifications. In total, we identified 276 proteins with a positive log2 fold-change in SARS-CoV-2 purified samples, relative to RMRP purified samples (Supplementary Table 1 ). Of these, 57 proteins were enriched with high statistical significance (P < 0.05, two-tailed moderated t-test), which we subsequently defined as the set of core SARS-CoV-2 interacting proteins ( Figure 1c ).

Additionally, we also defined an expanded SARS-CoV-2 RNA interactome using a relaxed false discovery rate (FDR) of less than 20% (Figure 1c ).

The expanded SARS-CoV-2 RNA interactome encompasses 104 human proteins and includes 13 SARS-CoV-2-encoded proteins. The vast majority of the human RNA interactome proteins (96 proteins, 92%) have previously been identified in high-throughput studies aimed at capturing proteins that crosslink to RNA 32 (Supplementary Table 2 ). Table 2 ). Next, we compared our direct SARS-CoV-2 RNA interactome with the group of proteins found to directly or indirectly associate with the RNA genomes of Dengue and Zika viruses 19 . Sixty-six proteins (63%) of the expanded SARS-CoV-2 RNA interactome were also found to associate with the Dengue and Zika virus RNAs, while 35 proteins (34%) were unique SARS-CoV-2 binders (Figure 2a ).

Since coronaviruses are known to form replication/transcription complexes (RTCs), we also compared the expanded SARS-CoV-2 RNA interactome to the protein content of murine coronavirus RTCs 33 and found 64 shared proteins (Supplementary Table 2 ).

Finally, a recent study identified protein-protein interactors of recombinant SARS-CoV-2 proteins expressed in uninfected human cells 11 . Only 10 of the 334 human proteins that were found to interact with viral proteins in uninfected cells, also bound directly to viral Table 2 ). Beyond the well-documented impact of viral infection on host cell gene expression [5] [6] [7] , which may lead to remodeling of observed interactions, these results further highlight the importance of discriminating between proteinprotein and RNA-protein interactions when dissecting the biology of SARS-CoV-2.

While our SARS-CoV-2 RNA interactome captured RNA-protein interactions not found in other RNA viruses or uninfected cells, many identified proteins were shared among these datasets, confirming the validity of our experimental approach. Further, the RNAprotein interactions only observed in SARS-CoV-2 RAP-MS experiments, may point towards unique aspects of SARS-CoV-2 biology.

To gain insights into the biological functions of SARS-CoV-2-bound proteins, we performed a hypergeometric gene ontology (GO) enrichment analysis 34 Table 3 ). Consistent with the enrichment of these GO-terms, the importance of mRNA translation at the endoplasmatic reticulum (ER) membrane is well-established for the replicative cycle of coronaviruses 35, 36 .

Further, nonsense-mediated RNA decay (NMD) was recently described as an intrinsic antiviral mechanism targeting coronavirus RNAs 37 .

In agreement with the crucial role of mRNA translation and its regulation, the expanded SARS-CoV-2 RNA interactome included 19 ribosomal proteins and 12 translation factors. Among translation factors, EIF4G1, which together with EIF4A and EIF4E, makes up the EIF4F complex, and EIF4B have been reported as targets of mammalian target of rapamycin (mTOR) signaling [38] [39] [40] . EIF4B plays a critical role in recruiting the 40S ribosomal subunit to mRNA, and both the PI3K/mTOR and mitogen-activated protein kinase (MAPK) pathways target EIF4B to control its phosphorylation status and activity 40 . PI3K/mTOR inhibition has further been demonstrated to suppress SARS-CoV-2 replication in human cells 8, 10 .

In order to systematically examine the connectivity of the identified SARS-CoV-2 binding proteins and their potential relationship to virus-associated biological processes, we We next integrated known drug-target interactions within this network and identified 23 SARS-CoV-2-bound human proteins that can be targeted with existing compounds, including PPIA, ANXA1, CFL1, and EGFR ( Figure 2e) . Notably, the core SARS-CoV-2 RNA interactome member ANXA1 plays an essential role as an effector of glucocorticoid-mediated regulation of the innate immune response. ANXA1 is a known target of dexamethasone, which according to newly emerging clinical trial data appears to show efficacy as a treatment for COVID-19 41 .

To complement our RNA-protein interactome and gain deeper insights into host response pathways activated upon SARS-CoV-2 infection, we globally measured protein abundance To identify cellular pathways modulated upon infection, we performed gene set enrichment analysis (GSEA) using our proteome abundance measurements (see Methods).

Among the most significantly enriched hallmark gene sets were "TGFβ signaling" (adj. p = 0.0048), "TNFα signaling via NFκB" (adj. p = 0.0018), "Interferon (IFN)-γ response" (adj. p = 0.0048), "IL-6 JAK STAT3 signaling" (adj. p = 0.0054), and "IFN-α response" (adj. p = 0.0077; Table 6 ). Newly emerging evidence indicates that these pathways are indeed highly relevant in the context of SARS-CoV-2 infections 5,7-10, 42 . Notably, inhibition of growth factor signaling through the MAPK pathway, which responds to and controls the production of pro-inflammatory cytokines, including TNFα and IL-6, was recently shown to modulate SARS-CoV-2 replication in infected cells [8] [9] [10] .

In agreement with recent transcriptome studies 5-7 , our proteome data suggest activation of interferon signaling upon SARS-CoV-2 infection. Among interferon-related genes, we observed significant upregulation of several major components of IFN signaling pathways, including STAT1 and IRF9, which together with STAT2 make up the ISGF3 complex, their upstream components TYK2 and JAK1, as well as their downstream targets IFIT1, IFIT3, IFITM3, OAS2, and ISG15 ( Figure 3a ). Other strongly upregulated IFN-related genes include BST2, SP110, UBE2L6, ADAR, TGIF1, BRD3, and IFI30 (Supplementary 71 , and YWHAZ 72 . In conclusion, our global protein abundance measurements verify the induction of an appropriate host response in SARS-CoV-2 infected HuH-7 cells and further support an essential role for IFN and MAPK signaling in SARS-CoV-2 infections.

To examine the connectivity between SARS-CoV-2 RNA interactome proteins and proteins Table 7 ; see Methods). In addition to a cluster of ribosomal proteins and translation factors, we observed many connections between canonical RNA-binding proteins linked to viral infections (including MOV10, YBX1, HNRNPA1, PUM1, DDX3X, HDLBP, SND1, and CFL1) 73 and regulated host factors in this network. Notably, we also observed non-classical RNA-binding proteins such as ANXA1 or ACTR2 among highly-connected proteins in this network. ACTR2 has been identified as a host factor for respiratory syncytial virus (RSV) infection and is involved in cytoskeleton reorganization and filipodia formation, which is thought to promote virus spreading 74, 75 . This observation is consistent with recent work that demonstrated that SARS-CoV-2 infection induced a dramatic increase in filopodia, and viral particles localized to and emerged from these protrusions 9 .

Taken together, this network analysis suggests extensive connectivity between SARS-CoV-2-binders and dynamically regulated host cell proteins, which connect interactome proteins to emerging SARS-CoV-2 biology and may provide a map for identifying regulatory hubs in SARS-CoV-2 infections.

We next focused our analysis on SARS-CoV-2 RNA interactome members that were strongly upregulated in our proteome dataset following viral infection. Among the 104 proteins in the expanded SARS-CoV-2 RNA interactome, 22 human proteins were significantly induced upon infection. In this group, PPIA, RAB6D, MSI2, SCFD1, and CNBP were the most strongly upregulated candidates (log2 fold-change > 2.5) and all of these proteins also showed robust enrichment in RAP-MS experiments (log2 fold-change > 1, P <0.1). Notably, PPIA was previously shown to be essential for SARS-CoV-1 replication 76, 77 , suggesting that the group of virally induced human SARS-CoV-2 interactors may constitute important candidates for a functional investigation.

To further dissect these candidates and validate our RAP-MS results, we first focused on CNBP, which represents the most significantly enriched protein in SARS-CoV-2 purifications and was also induced upon infection. CNBP translocates to the nucleus upon immune stimulation and associates with an immune stimulatory DNA oligonucleotide, suggesting that it may be involved in foreign nucleic acid-sensing pathways 78 . Two recent studies showed that CNBP activates the innate immune response and induces the expression of several pro-inflammatory cytokines, including Il-6 and Cxcl10 in vivo 78, 79 . While our proteomics experiments did not yield peptides for these two CNBP-regulated cytokines, recent transcriptome studies demonstrated that IL-6 and CXCL10 were indeed induced in SARS-CoV-2 infected human cells [5] [6] [7] . In light of these results, we speculated that in addition to its previously described role in foreign DNA sensing, CNBP might recognize the SARS-CoV-2 RNA through direct binding, which may be linked to its role in transcriptionally activating specific innate immune genes.

To corroborate the physical engagement of the SARS-CoV-2 RNA by CNBP in infected human cells, we combined RNA antisense purification (RAP) with enhanced crosslinking and immunoprecipitation (eCLIP) 80 Given the enrichment of various proteins linked to the IFN response in our SARS-CoV-2 RNA interactome, we speculated that inhibiting a central regulator of antiviral RNA sensing might make HuH-7 cells more permissive to SARS-CoV-2 replication. We noticed that several SARS-CoV-2 interacting proteins engage in well-documented physical interactions with TANK-binding kinase (TBK1). Of particular note, DDX3X is a TBK1 substrate and direct binding partner that synergizes with TBK1 in regulating IFN production 57 .

Similarly, ANXA1 regulates type I IFN signaling through its physical association with TBK1 65 , while STRAP interacts with TBK1 and IRF3 to promote IFN-β production 63 In addition to the induction of an IFN response, we provide evidence for the activation of MAPK and TNFα signaling pathways upon SARS-CoV-2 infection, which adds to growing evidence that these pathways are regulated upon SARS-CoV-2 infection and may be exploited to manipulate viral replication [8] [9] [10] . In the context of EGF/MAPK signaling, it is intriguing that we detected EGFR together with several RAS-related proteins, including RAB6D, RAB6A and RAB7A in our expanded RNA interactome. RAS-related proteins were also identified as protein-protein interaction partners of viral proteins in human cells, including RAB7A, which we find to bind to the SARS-CoV-2 RNA in this study. Together, these data may point at physical interactions between SARS-CoV-2 components and cellular pathways that are perturbed upon viral infection.

Beyond characterizing the SARS-CoV-2 RNA interactome, our work uncovered a subset of SARS-CoV-2 binders that are induced upon viral infection. Importantly, this group of proteins includes a previously identified host dependency factor essential for coronavirus replication (PPIA), suggesting that our approach indeed recovers relevant regulators of coronavirus biology. We characterized two of these proteins in greater detail. First, we demonstrated that CNBP directly interacts with the SARS-CoV-2 RNA, which expands its previously known role in the context of foreign DNA sensing. It is tempting to speculate that recognition of the SARS-CoV-2 RNA by CNBP might contribute to establishing a specific cytokine response in infected cells. Intriguingly, the known CNBP-dependent cytokine expression signature 78,79 bears similarity to gene expression changes in SARS-CoV-2 infected cells [5] [6] [7] , which warrants future investigations. In addition to CNBP, we provide evidence that the interferon-induced protein RYDEN interacts with SARS-CoV-2 RNA and suppresses programmed ribosome frameshifting in a SARS-CoV-2 translational reporter.

Together, these insights illuminate how host cell proteins directly engage foreign RNA molecules during infection and contribute to different host defense strategies. Our characterization of the SARS-CoV-2 RNA interactome is a valuable resource that provides an RNA-centric perspective on SARS-CoV-2 infections and will help to decode both viral RNA metabolism as well as host defense mechanisms. As demonstrated in our study, the identified interactions recover known coronavirus biology and uncover opportunities for targeting proteins or pathways linked to the SARS-CoV-2 RNA interactome in order to control viral replication. We believe that our findings and strategies provide a general roadmap for dissecting the biology of RNA viruses and the interactions between hosts and pathogens at the molecular level.

We maintained HuH-7 cells in DMEM media (Thermo Fisher Scientific) supplemented with 10% heat-inactivated FBS (Thermo Fisher Scientific), and 100 units/ml streptomycin and 100 mg/ml penicillin. Cells were grown at 37°C and 5% CO 2 atmosphere.

We used a previously described patient-derived SARS-CoV-2 isolate 84 that was propagated on HuH-7 cells and characterized by RNA sequencing. Viral loads were frequently determined by RT-qPCR. In brief: viral RNA was extracted with a MagNA Pure 24 system (Roche, Germany) and quantified with the LightMix Assay SARS-CoV-2 RdRP RTqPCR assay kit (TIB MOLBIOL, Germany) and the RNA Process Control kit (Roche). Viral titers were determined by immunofluorescence.

RAP-MS was carried out as previously described 13 Lysates were then collected and flash-frozen in liquid nitrogen for storage at -80°C. All subsequent steps were carried out at previously described 13 .

RAP-captured proteins were resuspended in 40 μL of 8 M urea in 50 mM Tris-HCl, followed by reduction with 4 mM DTT for 30 minutes at room temperature and alkylation with 10 mM IAA for 45 minutes at room temperature in the dark. All six samples were then digested with 0.1 μg Lys-C for 2 hours, followed by a reduction of the urea concentration to <2 M and continued digestion with 0.5 μg trypsin overnight. Reactions were quenched with formic acid at a final concentration of 5% and then desalted by reverse phase C18 stage tips as described previously 87 

For proteome measurements, we expanded HuH-7 cells to two 10cm tissue culture plates per replicate. Cells were infected with an MOI of 10 and incubated for 24 hours before harvesting. Three process replicates of infected and non-infected cell line samples were generated. Cells were lysed in 8 M Urea, 75 mM NaCl, 50 mM Tris pH 8.0, 1mM EDTA, 2 µg/ml Aprotinin, 10 μg/ml Leupeptin, 1 mM PMSF, 10 mM NaF, PIC2 (Sigma Aldrich), PIC3 (Sigma Aldrich) and 10mM Sodium Butyrate. Benzonase was added to digest nucleic acids and DNA was sheared using a probe sonicator (10% amplitude, 0.7s on, 2.3s off, 6 min 15s total). Cell debris was removed by centrifugation and lysates were flash-frozen for storage.

All samples were prepared for MS analysis using an optimized workflow as described previously 88 . Briefly, lysed samples were reduced, alkylated, and digested by LysC for 2 hours, followed by overnight digestion with trypsin. Digestions were quenched with formic acid and all peptide samples were desalted using reverse phase C18 SepPak cartridges. Samples were then quantified by the Pierce BCA Protein Assay and measured into 500 μg aliquots for isobaric labeling. Peptides were isobarically labeled with TMT6 following the reduced TMT protocol 89 Carbamidomethylation and TMT labeling at Lysine (with and without labeling at N-terminus) were set as fixed modifications, while variable modifications included acetylation of protein Ntermini, oxidized methionine, deamidation of asparagine, and pyro-glutamic acid at peptide N-terminal glutamine. Peptide spectrum match score thresholding was optimized to achieve a target-decoy false discovery rate (FDR) of 1.2% for validation of spectra. Peptide level auto-validation was followed by protein polishing with FDR of 0% at protein level and minimum score of 13. The Spectrum Mill generated proteome level export from the RAP-MS and proteome datasets filtered for human proteins identified by two or more distinct peptides, SARS-CoV-2 proteins and unannotated virus ORFs were used for further statistical analyses. Five of the SARS-CoV-2 non-structural proteins (NSP6, NSP15, NSP16, NSP9, and NSP1) identified by a single, highly scoring distinct peptide were kept in the dataset. Keratins were excluded from RAP-MS data. Protein quantification was achieved by taking the ratio of TMT reporter ion for each sample/channel over the median of all 6 channels. Moderated two-sample t-test was applied to compare SARS-CoV-2 and RMRP samples after mean normalization and SARS-CoV-2 infected and non-infected samples after median-MAD normalization of RAP-MS and proteome datasets, respectively. Benjamini-Hochberg corrected p-value threshold of 0.05 was used to asses significantly regulated proteins in each of the datasets.

Following the described RAP-MS procedure, we saved 5% of benzonase eluted proteins for western blot analysis. We added NuPAGE LDS Sample Buffer (Thermo Fisher Scientifc) and incubated samples for 3 min incubation at 95°C. Proteins were resolved by SDS-PAGE using 

To capture RNA sequences covalently crosslinked to proteins purified with RAP-MS, we carried out RNA antisense purifications as described above. Following our pilot RAP-MS experiment (see Supplementary Figure 1a 

We seeded 10 5 HuH-7 cells per well of a 24-well plate. The next day, growth medium was replaced by medium containing Cyclosporin A (Sigma-Aldrich, SML1018) or BX-795 hydrochloride (Sigma-Aldrich, SML0694) at indicated concentrations 2h prior to infection. Cells were infected with SARS-CoV-2 at MOI 0.5 PFU/cell by adding virus on top of the inhibitor-containing media.

We seeded 2*10 4 HuH-7 cells per well of a 96-well plate. The next day, the growth medium was replaced by medium containing Cyclosporin A (Sigma-Aldrich, SML1018) or BX-795 hydrochloride (Sigma-Aldrich, SML0694) at indicated concentrations. Cells were incubated for 72 h, and cell viability was assessed using the Cell-Titer-Glo reagent (Promega) according to the manufacturer's instructions. providing a direct readout of ribosomal frameshifting efficiency. Accordingly, frameshifting efficiency was calculated using the ratio of mCherry to EGFP observed with the frameshifting reporter construct (Supplementary Figure 3a) , relative to the mCherry/EGFP ratio observed with the control construct (Supplementary Figure 3a) .

To establish protein-protein interactions for the proteins identified from the MS experiments, we utilized STRING v11 93 . Specifically, for the RAP-MS network (Figure 2 ), we seeded all proteins detected with an adj. P < 0.2 and positive logFC from the moderated t-test between SARS-CoV2 purifications and RMRP purifications. Edges between interacting proteins were included for those above a combined interaction score of 550. To generate the combined RAP-MS and proteome MS network, we seeded nodes where the adj. P < 0.05 for either of the assays. Edges between RAP-MS nodes and proteome MS nodes were included for combined interaction scores exceeding 700.

We performed a hypergeometric Gene Ontology (GO) enrichment analysis for the expanded SARS-CoV-2 interactome proteins using the Database for Annotation, Visualization and Integrated Discovery (DAVID) tool (https://david.ncifcrf.gov/tools.jsp) and applying default settings.

We performed Gene Set Enrichment Analysis (GSEA) for proteome experiments with the clusterProfilter R package 94 utilizing the Hallmark and C5-Biological Processes gene sets available through MSigDB 95 . Genes were ranked based on the product of the log2 fold change and the log10 moderated t-test P value between the SARS-CoV2 treatment and mock treatments.

Paired-end sequencing reads from (i) eCLIP experiments, or (ii) sequencing of crosslinked RNA fragments following RAP-MS, were trimmed using a custom python script that simultaneously identified the umi-molecular identifier (UMI) associated with each read. These trimmed reads were then aligned to the SARS-Cov2 reference genome (NC_045512.2 contig) using bwa 96 . Next, we removed PCR duplicates using the UMI-aware deduplication functionality in Picard's MarkDuplicates. Finally, enriched regions of binding were identified using the MACS2 97 to model the fold change between per-million fragment normalized counts (--SPMR) of the treated and control. Visualizations of the region were rendered from the PCR-deduplicated .bam files using the Integrative Genome Visualization (IGV) Browser.

Custom computer code and an interactive rendering of our protein-protein association network of SARS-CoV-2 RNA interacting proteins is publicly accessibly at: https://munschauerlab.github.io/SCoV2-proteome-atlas/. 

Clinical Characteristics of Coronavirus Disease 2019 in China

The trinity of COVID-19: immunity, inflammation and intervention

Intracellular sensing of viral genomes and viral evasion

Viral evasion of intracellular DNA and RNA sensing

Bulk and single-cell gene expression profiling of SARS-CoV-2 infected human cell lines identifies molecular targets for therapeutic intervention

Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19

Single-cell longitudinal analysis of SARS-CoV-2 infection in human bronchial epithelial cells

Growth factor receptor signaling inhibition prevents SARS-CoV-2 replication

The Global Phosphorylation Landscape of SARS-CoV-2 Infection

Multi-level proteomics reveals host-perturbation strategies of SARS-CoV-2 and SARS-CoV

A SARS-CoV-2 protein interaction map reveals targets for drug repurposing

The Xist lncRNA interacts directly with SHARP to silence transcription through HDAC3

The NORAD lncRNA assembles a topoisomerase complex critical for genome stability

Systematic Discovery of Xist RNA Binding Proteins

Methods to study RNA-protein interactions

Enhanced isolation of SARS-CoV-2 by TMPRSS2-expressing cells

Highly infectious SARS-CoV pseudotyped virus reveals the cell tropism and its correlation with receptor expression

The RNA-binding proteomes from yeast to man harbour conserved enigmRBPs

An RNA-centric dissection of host complexes controlling flavivirus infection

<p>Long Noncoding RNA RMRP Suppresses the Tumorigenesis of Hepatocellular Carcinoma Through Targeting microRNA-766</p>

The Nonstructural Proteins Directing Coronavirus RNA Synthesis and Processing

Crystal structure of Nsp15 endoribonuclease NendoU from SARS CoV 2

Structure of replicating SARS-CoV-2 polymerase

Structural Basis of RNA Cap Modification by SARS-CoV-2 Coronavirus

The nsp9 Replicase Protein of SARS-Coronavirus

Proteomics analysis unravels the functional repertoire of coronavirus nonstructural protein 3

The 3a Accessory Protein of SARS Coronavirus Specifically Interacts with the 5'UTR of Its Genomic RNA, Using a Unique 75 Amino Acid Interaction Domain †

Coronavirus genomic RNA packaging

Cell entry mechanisms of SARS-CoV-2

Severe acute respiratory syndrome coronavirus nonstructural proteins 3, 4, and 6 induce doublemembrane vesicles

The coding capacity of SARS-CoV-2

RBPbase v0.1.8 alpha: A comprehensive database of eukaryotic RNA-binding proteins (RBP) with their RBP annotations

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

Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources

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

Modification of intracellular membrane structures for virus replication

Interplay between coronavirus, a cytoplasmic RNA virus, and nonsense-mediated mRNA decay pathway

Targeting the eIF4F translation initiation complex: a critical nexus for cancer development

eIF4GI links nutrient sensing by mTOR to cell proliferation and inhibition of autophagy

The mTOR/PI3K and MAPK pathways converge on eIF4B to control its phosphorylation and activity

Effect of Dexamethasone in Hospitalized Patients with COVID-19: Preliminary Report

Genome-wide CRISPR screen reveals host genes that regulate SARS-CoV-2 infection

A novel function of human Pumilio proteins in cytoplasmic sensing of viral infection

YB-1 suppression induces STAT3 proteolysis and sensitizes renal cancer to interferon-α

Human interactome of the influenza B virus NS1 protein

Heterotrimeric GAIT complex drives transcript-selective translation inhibition in murine macrophages

G3BP1 and G3BP2 regulate translation of interferon-stimulated genes: IFITM1, IFITM2 and IFITM3 in the cancer cell line MCF7

G3BP2 and CAPRIN1 are required for translation of interferon stimulated mRNAs and are targeted by a dengue virus noncoding RNA

G3BP1 promotes DNA binding and activation of cGAS

The stress granule protein G3BP1 binds viral dsRNA and RIG-I to enhance interferon-β response

Interferon-dependent engagement of eukaryotic initiation factor 4B via S6 kinase (S6K)-and ribosomal protein S6K-mediated signals

P body-associated protein Mov10 inhibits HIV-1 replication at multiple stages

MOV10 Provides Antiviral Activity against RNA Viruses by Enhancing RIG-I-MAVS-Independent IFN Induction

Host Protein Moloney Leukemia Virus 10 (MOV10) Acts as a Restriction Factor of Influenza A Virus by Inhibiting the Nuclear Import of the Viral Nucleoprotein

DEAD/H BOX 3 (DDX3) helicase binds the RIG-I adaptor IPS-1 to up-regulate IFN-beta-inducing potential

DDX3 directly regulates TRAF3 ubiquitination and acts as a scaffold to co-ordinate assembly of signalling complexes downstream from MAVS

The DEAD-box helicase DDX3X is a critical component of the TANKbinding kinase 1-dependent innate immune response

Viral targeting of DEAD box protein 3 reveals its role in TBK1/IKKepsilon-mediated IRF activation

LSm14A is a processing body-associated sensor of viral nucleic acids that initiates cellular antiviral response in the early phase of viral infection

LSM14A inhibits porcine reproductive and respiratory syndrome virus (PRRSV) replication by activating IFN-β signaling pathway in Marc-145

Characterization of RyDEN (C19orf66) as an Interferon-Stimulated Cellular Inhibitor against Dengue Virus Replication

Regulation of HIV-1 Gag-Pol Expression by Shiftless, an Inhibitor of Programmed -1 Ribosomal Frameshifting

STRAP positively regulates TLR3-triggered signaling pathway

Annexin-A1 promotes RIG-I-dependent signaling and apoptosis via regulation of the IRF3-IFNAR-STAT1-IFIT1 pathway in A549 lung epithelial cells

Annexin-A1 Regulates TLR-Mediated IFN-β Production through an Interaction with TANK-Binding Kinase 1

DDX1, DDX21, and DHX36 helicases form a complex with the adaptor molecule TRIF to sense dsRNA in dendritic cells

Cellular RNA Helicase DDX1 Is Involved in Transmissible Gastroenteritis Virus nsp14-Induced Interferon-Beta Production

The cellular RNA helicase DDX1 interacts with coronavirus nonstructural protein 14 and enhances viral replication

PCBP2 enhances the antiviral activity of IFN-α against HCV by stabilizing the mRNA of STAT1 and STAT2

NLRX1 Mediates MAVS Degradation To Attenuate the Hepatitis C Virus-Induced Innate Immune Response through PCBP2

Nuclear hnRNPA2B1 initiates and amplifies the innate immune response to DNA viruses

14-3-3ζ Regulates Immune Response through Stat3 Signaling in Oral Squamous Cell Carcinoma

Unconventional RNA-binding proteins step into the virus-host battlefront

A novel host factor for human respiratory syncytial virus

Actin-Related Protein 2 (ARP2) and Virus-Induced Filopodia Facilitate Human Respiratory Syncytial Virus Spread

Cyclosporin A inhibits the replication of diverse coronaviruses

The SARS-coronavirus-host interactome: identification of cyclophilins as target for pan-coronavirus inhibitors

CNBP controls IL-12 gene transcription and Th1 immunity

CNBP acts as a key transcriptional regulator of sustained expression of interleukin-6

Robust transcriptome-wide discovery of RNA-binding protein binding sites with enhanced CLIP (eCLIP)

La-related protein 4 binds poly(A), interacts with the poly(A)-binding protein MLLE domain via a variant PAM2w motif, and can promote mRNA stability

Cyclophilin A and viral infections

Evidence for host-dependent RNA editing in the transcriptome of SARS-CoV-2

The serotonin reuptake inhibitor Fluoxetine inhibits SARS-CoV-2

RNA-RNA interactions enable specific targeting of noncoding RNAs to nascent Pre-mRNAs and chromatin sites

The Xist lncRNA Exploits Three-Dimensional Genome Architecture to Spread Across the X Chromosome

Cell-Surface Proteomic Profiling in the Fly Brain Uncovers Wiring Regulators

Reproducible workflow for multiplexed deep-scale proteome and phosphoproteome analysis of tumor tissues by liquid chromatography-mass spectrometry

TMT Labeling for the Masses: A Robust and Cost-efficient, In-solution Labeling Approach

A new coronavirus associated with human respiratory disease in China

Higher-Order Inter-chromosomal Hubs Shape 3D Genome Organization in the Nucleus

The Human RNA-Binding Proteome and Its Dynamics during Translational Arrest

STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets

clusterProfiler: an R package for comparing biological themes among gene clusters

Molecular signatures database (MSigDB) 3.0

Fast and accurate short read alignment with Burrows-Wheeler transform

Model-based analysis of ChIP-Seq (MACS)

gene sets are shown; full table displaying additional enriched gene sets is provided in Supplementary Figure 2b. c, Protein-protein association network of core SARS

RNA interactome proteins and their connections to differentially regulated proteins in SARS-CoV-2 infected cells, based on curated interactions in STRING v11 93

The authors declare no competing interests.

Mathias Munschauer (mathias.munschauer@helmholtz-hiri.de) (lead contact),