key: cord-0736204-2j6y8wlm
authors: Nabeel-Shah, Syed; Lee, Hyunmin; Ahmed, Nujhat; Marcon, Edyta; Farhangmehr, Shaghayegh; Pu, Shuye; Burke, Giovanni L.; Ashraf, Kanwal; Wei, Hong; Zhong, Guoqing; Tang, Hua; Yang, Jianyi; Blencowe, Benjamin J.; Zhang, Zhaolei; Greenblatt, Jack F.
title: SARS-CoV-2 Nucleocapsid protein attenuates stress granule formation and alters gene expression via direct interaction with host mRNAs
date: 2020-10-23
journal: bioRxiv
DOI: 10.1101/2020.10.23.342113
sha: 8bb0ded1d6dea31d207bfe1a213e9a7ba7694347
doc_id: 736204
cord_uid: 2j6y8wlm

The COVID-19 pandemic has caused over one million deaths thus far. There is an urgent need for the development of specific viral therapeutics and a vaccine. SARS-CoV-2 nucleocapsid (N) protein is highly expressed upon infection and is essential for viral replication, making it a promising target for both antiviral drug and vaccine development. Here, starting from a functional proteomics workflow, we initially catalogued the protein-protein interactions of 21 SARS-CoV-2 proteins in HEK293 cells, finding that the stress granule resident proteins G3BP1 and G3BP2 copurify with N with high specificity. We demonstrate that N protein expression in human cells sequesters G3BP1 and G3BP2 through its physical interaction with these proteins, attenuating stress granule (SG) formation. The ectopic expression of G3BP1 in N-expressing cells was sufficient to reverse this phenotype. Since N is an RNA-binding protein, we performed iCLIP-sequencing experiments in cells, with or without exposure to oxidative stress, to identify the host RNAs targeted by N. Our results indicate that SARS-CoV-2 N protein binds directly to thousands of mRNAs under both conditions. Like the G3BPs stress granule proteins, N was found to predominantly bind its target mRNAs in their 3’UTRs. RNA sequencing experiments indicated that expression of N results in wide-spread gene expression changes in both unstressed and oxidatively stressed cells. We suggest that N regulates host gene expression by both attenuating stress granules and binding directly to target mRNAs.

Here we show that the SARS-CoV-2 N protein physically interacts with G3BP1/2 and attenuates SG formation. We found that, like stress granule proteins G3BP1/2, N protein binds directly to cellular mRNAs, with a preference for 3'UTRs. Furthermore, we report that expression of N alters the expression profile of genes implicated in major biological processes and cellular pathways. We suggest that N protein alters host gene expression levels via both SG attenuation and direct interaction with mRNAs of many genes.

To examine the potential impacts of viral proteins on host cell metabolism, we generated HEK293 cell lines expressing EGFP-tagged variants of 21 full-length SARS-Cov-2 genes (Supplemental Table S1 ). Each of the EGFP-tagged proteins was subjected to affinity purification with anti-GFP antibodies followed by Orbitrap-based precision mass spectrometry analysis (AP-MS). To remove any possible indirect associations mediated by DNA or RNA, cell extracts were treated with a promiscuous nuclease (Benzonase) prior to the affinity pull-downs. We scored specific protein interactions against GFP control purifications using SAINTexpress analysis 26 with a statistical cutoff of ≤ 1% false discovery rate [FDR] . In total, we identified 647 protein-protein interactions (PPIs) involving 277 unique cellular proteins (Extended Figure 1 ; Supplemental Table S1 ). We performed KEGG enrichment analysis using the purified interaction partners and identified major cellular pathways, including RNA transport, protein processing in the endoplasmic reticulum, the proteosome and oxidative phosphorylation (P-value cut-off 0.05 (FDR)) (Extended Figure 2A , B).

In addition, Huntington, Parkinson's and Alzheimer's disease related pathways were significantly enriched among the identified interaction partners (P-value cut-off 0.05 (FDR)) (Extended Figure   2B ). We next compared our protein interaction data with two previously published AP-MS studies that were also performed in HEK293 cells 13, 14 . We observed that there is little interaction partner overlap among the different reports, highlighting the variability of the experimental and/or statistical analyses employed in the various studies to catalogue the virus-host PPIs (see discussion) ( Figure 1A) . Remarkably, the only shared interaction partners among all three studies were the stress granule nucleating proteins G3BP1 and G3BP2. Considering their essential role in stress granule formation 17, 18 , we next focused our attention on examining the functional significance of G3BP1 and G3BP2 interaction with the SARS-CoV-2 proteins.

Our AP-MS analysis indicated that, among the 21 examined SARS-CoV-2 proteins, G3BP1/2 copurified exclusively with the Nucleocapsid N protein ( Figure 1B; Extended Figure 1 ). We confirmed the interaction between N and G3BP1 by co-immunoprecipitation (co-IP) experiments using whole cell extracts (WCEs) prepared from cells expressing GFP-N and FLAG-G3BP1 ( Figure 1C ). Furthermore, the interaction was resistant to nuclease treatment indicating that it is not mediated by RNA ( Figure 1C ). The N protein NTD, which contains a pocket for binding viral RNA, is highly conserved across the coronaviruses and has been described as a potentially druggable target 8 . Recently, NMR structural studies identified several arginine (R) residues, including R92, R107, and R149, that directly contact the RNA 27 . Since crystal structures for both the SARS-CoV-2 N protein NTD and G3BP1 are available 27,28 , we performed protein docking studies to identify the residues on the two proteins that might come directly into contact. Our analyses predict a highly stable N-G3BP1 complex structure ( Figure 1D ), further supporting our AP-MS and co-IP analyses. We found that residues 90-96 (magenta spheres in Figure 1D ) of G3BP1 (blue cartoon) potentially interact with residues 130-134 (green spheres) of the N protein (red cartoon) ( Figure 1D ). We conclude, therefore, that the SG resident proteins G3BP1/2 interact physically with the viral N protein and that this interaction is highly specific, as no other SARS-CoV-2 protein was observed to pull down G3BPs.

To begin elucidating the role of N in the context of SG formation, we performed AP-MS analysis using G3BP1 as the bait. SAINTexpress analysis indicated that, in addition to G3BP2, several SG nucleator proteins, including CAPRIN1, USP10, ATXN2l and NUFIP2, co-purify with G3BP1 as high-confidence (FDR<0.01) interaction partners (Figure 2A ). These high-confidence copurifying proteins were previously shown to function in SG formation upon exposure to stress 18, 20, 29 . Notably, the so-called 'essential' SG resident proteins that we identified as G3BP1 interaction partners were specifically depleted from the SARS-CoV-2 N interactome (compare Figures 1B and 2A) . While further studies are underway to directly examine any remodelling of the G3BP1/2 PPIs upon N expression, our current analysis suggests that N might function to sequester G3BP1/2 away from interacting SG nucleating proteins, thus attenuating the SG formation.

To test the hypothesis that N attenuates SG formation, we performed immunofluorescence studies.

Cells expressing either GFP-N or GFP alone were subjected to oxidative stress by sodium arsenite (NaAsO2) treatment for one hour. As identified by G3BP1 staining (Figure 2B ), SG punctae were readily observable in cells expressing both GFP and GFP-N Remarkably, we observed that the expression of N significantly reduced the number of observable SGs in cells subjected to NaAsO2 treatment ( Figure 2B ). Furthermore, we also observed that the N protein itself localized to the remaining SGs ( Figure 2B ). These data suggest that the expression of N negatively impacts SG formation upon exposure to stress.

Next, we examined whether this phenotype could be rescued through the overexpression of G3BP1 in N-expressing cells. N-expressing cells were transiently transfected with G3BP1 constructs followed by the induction of oxidative stress using NaAsO2. Consistent with the idea that N sequesters away G3BP1 from its interacting SG nucleators, we observed that the overexpression of G3BP1 rescued the SG attenuation phenotype ( Figure 2B ). As expected, N localized to the SGs in G3BP1 overexpressing cells. Overall, we conclude that N expression inhibits the formation of SGs in the host cells, possibly by sequestering away G3BPs from their interacting SG nucleating proteins.

Since N protein contains RNA-binding domains 8 , we also examined whether N interacts directly with host mRNAs. We initially performed crosslinking and immunoprecipitation (CLIP) followed by gel electrophoresis and autoradiography to detect crosslinked RNA. We observed that N cross-links robustly to RNA in vivo upon exposure to UV ( Figure 3A ). In contrast, GFP alone did not yield any observable radioactive signal, indicating that N directly binds host RNAs ( Figure 3A ). Next, to identify host mRNAs bound by N, we carried out individual-nucleotide resolution UV crosslinking and immunoprecipitation (iCLIP) followed by high throughput sequencing (iCLIP-seq) experiments in biological replicates along with size-matched inputs (SMI) as controls.

Through peak calling in comparison with the SMI controls, we identified >30,000 high confidence peaks encompassing ~4500 unique human protein-coding genes (Extended Figure 3A ;

Supplemental Table S2 ). Gene ontology (GO) enrichment analysis was carried out using these Nbound human genes. Major biological processes related to post-transcriptional regulation, including mRNA processing, RNA catabolic processes, RNA transport, post-transcription regulation of gene expression, and translation regulation were significantly enriched (FDR 0.05) (Extended Figure 3B ). KEGG enrichment analysis indicated that N-bound genes were enriched for pathways that included protein processing in endoplasmic reticulum, ribosome and cell cycle (FDR 0.05) (Extended Figure 3B ). We next examined the iCLIP peak distribution across the N-bound transcripts. Remarkably, most peaks (77% of the total peaks) were found within the annotated 3' untranslated regions (UTRs) ( Figure 3B ). We also searched for enriched sequence motifs and found that U-rich sequence motifs were significantly enriched among the N-binding sites ( Figure 3C ; Extended Figure 3C ). Since N interacted with G3BP1 and G3BP2, we included their PAR-CLIP-seq data 30 for comparison. Our metagene analyses indicated that, similar to the RNA-binding profiles of G3BP1 and G3BP2, the SARS-CoV-2 N protein predominantly binds within the 3'UTRs of its target genes, consistent with its peak distribution ( Figure 3D ). Our analysis revealed that ~47% of the G3BP1 and G3BP2 targets were also bound by the N protein ( Figure 3E) . Furthermore, N's iCLIP-seq signal was enriched around the G3BP1 and G3BP2 binding sites on the RNAs ( Figure 3F ). These results reinforce the idea that SARS-CoV-2 N and G3BPs might function together in the infected cells. It is conceivable that N might reshape the G3BP1/2-bound transcriptome to the advantage of SARS-CoV-2. Further experiments are underway to test this possibility (experimental data being analyzed). Additionally, studies are ongoing to investigate whether N binds cooperatively in conjunction with G3BPs on shared target mRNAs.

To examine the effect of SARS-CoV-2 N on host gene expression, we performed RNA-seq analysis in HEK293 cells expressing GFP-N. Since N appears to have a role in SG formation, we also included RNA samples prepared from cells treated with NaAsO2. Cells expressing GFP alone were used as controls in these experiments. The RNA-seq replicates highly correlated with each other, indicating the reproducibility of our data (Extended Figure 4) . Differential expression analysis identified 4363 and 2942 genes that were significantly differentially expressed in Nexpressing cells that were untreated or NaAsO2-treated (Q<0.05), respectively ( Figure 4A ). Of the 4363 differential genes in the untreated samples 2207 were upregulated, whereas 2156 were downregulated (Supplemental data S3,4). In the NaAsO2-treated cells, 1765 and 1178 genes were up-and down-regulated, respectively. We found that ~82% of the differential genes in the NaAsO2 samples (2403/2942 genes) were the same as those that were also differentially expressed in the untreated cells. These results indicated that N affects a core set of genes under both conditions. Furthermore, after normalizing the NaAsO2 RNA-seq data against the untreated samples, we identified a small subset of genes significantly upregulated in the N-expressing cells under stress conditions (Q<0.05) (Supplemental data S5).

We next examined whether the N-affected host genes are enriched for any specific biological processes. GO enrichment analysis indicated that the upregulated genes were significantly enriched in biological processes related to cellular localization, intracellular protein transport, and cell cycle regulation (Q-value cut-off 0.05; Extended Figure 5A ). Furthermore, KEGG pathway analysis showed that cancer-related pathways, TGF-beta signalling, Hippo signalling, protein processing in ER and RNA transport were significantly enriched in the upregulated genes (Q-value cut-off 0.05) ( Figure 4B ). The downregulated genes, on the other hand, were enriched in pathways related to neurogenesis and nervous system development (Q-value cut-off 0.05; Extended Figure   5B ). These data suggest that the expression of N results in deregulation of many functionally important genes.

Considering that N interacts with G3BP1, which has been shown to enhance the stability and translation of its target mRNAs 15, 30, 31 , we correlated the effect of N binding to its target mRNA abundance. Our results indicate that mRNAs associated with N were generally stabilized in comparison to the non-targets, which were significantly decreased in their abundance upon NaAsO2 treatment ( Figure 4B ). G3BP1 and G3BP2 also showed similar trends in our analyses (Extended Figure 6 ), consistent with their shared targets with N. These data suggest that the RNAbinding of N might stabilize certain target mRNAs against the stress-induced degradation (see discussion). Further studies are underway to directly identify the N targets in NaAsO2 treated cells and examine how N might reshape the G3BP1-bound transcriptome.

Given that the ongoing COVID-19 pandemic has globally caused more than one million deaths (World health organization data; October 5, 2020), there is an urgent need to better understand the SARS-CoV-2 life cycle for effective antiviral drug development. In this study, we report the biological impact of the SARS-CoV-2 N protein on host cells. Starting from a proteomic workflow, we found that there is little overlap across studies that recently reported virus-host PPIs for SARS-CoV-2 proteins 13, 14 . This observation is perhaps not surprising due to varied experimental and statistical analyses employed across different studies. We suggest that, in the absence of additional evidence to support these conclusions, caution should be exercised in interpreting proteomics data for drug development Among the high-confidence interactors detected across three AP-MS studies were the stress granule resident proteins G3BP1 and G3BP2. SGs are typically formed upon viral infection, possibly as a cellular response to block viral replication 16 . However, viruses have evolved many counter-measures to combat host responses to viral infection and, indeed, many viruses have been reported to destabilize SGs upon infection 15, 16 . Certain viruses have been reported to hijack G3BP1 and G3BP2, inhibiting SG formation to the benefit of the virus. In such cases, depletion of G3BPs (or other SG components) reduced viral replication 15, 16 . For example, Zika virus hijacks G3BPs, to reduce SG formation and benefit viral replication 22 , and it was shown that depletion of G3BP1 indeed reduced Zika virus replication 22 . In contrast to Zika, however, G3BPs have been found to inhibit the replication of Sendai virus and vesicular stomatitis virus 32 . Here we showed that SARS-CoV-2 N protein interacts physically with G3BPs and that expression of N attenuates SG formation. It remains to be seen, however, whether or not G3BPs also play a role in SARS-CoV-2 replication. In this context, it is interesting to note that a recent study showed that G3BPs interact with SARS-CoV-2 RNA 33 .

Although N expression resulted in both up-and down-regulation of various host genes, N-bound mRNAs were more stable in comparison to non-targets. Previous reports have shown that certain adenovirus and hepatitis C virus (HCV) proteins stabilize host mRNAs via binding to AU-rich elements present within 3'UTRs of target transcripts 34, 35 . AU-rich elements (ARE), which are present in many proto-oncogenes, growth factor and cytokine mRNAs, target mRNAs for degradation [36] [37] [38] . The adenovirus protein, E4orf6, has been reported to stabilize the host AREcontaining mRNAs, and this stabilization was found to be necessary for its oncogenic activity 34 .

Similarly, NS5A of HCV was shown to directly bind to U-(or G-) rich elements within 3'UTRs of host mRNAs, resulting in their stabilization 35 . The NS5A target mRNAs were found to be highly enriched in regulators of cell growth, cell death and cancer 35 . Based on these findings, it has been suggested that virus-induced stabilization of host transcripts prevents cell death and promotes growth of virus-infected cells 35 . While SARS-CoV-2 is an acute respiratory virus, our finding that N protein recognises U-rich elements and may stabilize its targets via directly binding to 3'UTRs of host mRNAs is of particular interest given that the up-regulated genes were significantly enriched in cancer-related pathways. Although mechanistic details of N-mediated stabilization remain unknown, we found that N and G3BPs not only physically interact with each other but also have similar RNA-binding profiles with many shared targets. G3BPs have been shown to stabilize their target mRNAs in unstressed as well as stressed cells reviewed by 15 . We suggest that SARS-CoV-2 N and G3BPs bind to various target mRNAs, resulting in their stabilization.

It is worth noting that N-bound stabilized targets were also enriched in major cellular pathways, including "protein processing in ER" and "TGF-beta signalling". This finding is of interest, as viruses, including SARS-CoV, have been shown to cause ER stress and the induction of signalling pathways collectively known as the unfolded protein response (UPR) 39, 40 . However, coronaviruses are thought to have evolved the ability to subvert, or even exploit, certain aspects of the UPR and overcome protein translation shutdown 39 . Further studies will be required to investigate whether or not the stabilization and/or direct binding of N to ER-related mRNAs plays a role in the ER restructuring that is observed upon infection.

Further studies are underway to investigate the role of N upon oxidative stress, including how N protein might rewire the G3BP RNA-binding profile and what mechanisms N protein might use to reshape the host transcriptome. Based on the evidence presented here, we suggest that SARS-CoV-2 N protein affects host cell metabolism and gene expression via multiple pathways, including SG attenuation, sequestration of G3BPs, and direct binding to mRNAs of some functionally important genes. Considering that N is a promising target for drug and vaccine development, our findings might be of therapeutic interest. 

To perform Co-IP experiments, cell pellets were lysed in 1mL of lysis buffer (140 mM NaCl, 10 mM Tris pH 7.6-8.0, 1% Triton X-100, 0.1% sodium deoxy-cholate, 1 mM EDTA) containing protease inhibitors (Roche catalogue number 05892791001). Cell extracts were incubated with 75 units of Benzonase (Sigma E1014) for 30min in a cold room with end-to-end rotation. The cell lysates were cleared in a microcentrifuge at 15,000 g for 30 minutes at 4°C. The supernatant was transferred to a new tube and incubated with 1μg of GFP antibody for 4 hours to overnight, and subsequently 10 μl protein G Dyna beads were added and incubated for an additional 2 hours (Invitrogen catalogue number 10003D). The samples were washed three times with lysis buffer containing an additional 2% NP40 for 5 min each in a cold room with end-to-end rotation. The samples were then boiled in SDS gel sample buffer. Samples were resolved using 4-12% BisTris-PAGE and transferred to a PVDF membranes (Bio-Rad catalogue number 162-0177) using a Gel Transfer Cell (BioRad catalogue number 1703930). Primary antibodies were used at 1: 5000 dilution, and horseradish peroxidase-conjugated goat anti-mouse (Thermo Fisher 31430) or antirabbit secondary (Thermo Fisher 31460) antibodies were used at 1:10,000. Blots were developed using Pierce ECL Western Blotting Substrate (Thermo Scientific catalogue numbers 32106).

Individual nucleotide resolution UV crosslinking and immunoprecipitation (iCLIP) was performed as previously described 43 with the modifications as detailed in our previous report 44 

The AP-MS procedure for HEK293 cells was performed essentially as previously described 41, 45 .

Briefly, ∼20x10 6 cells were grown in two independent batches representing biological replicates.

After 24 h induction of protein expression using doxycycline, cells were harvested. Cell pellets were lysed in high-salt NP-40 lysis buffer (10 mM Tris-HCl pH 8.0, 420 mM NaCl, 0.1% NP-40, plus protease/phosphatase inhibitors) with three freeze-thaw cycles. The lysate was sonicated as described 41 . To remove genomic DNA and RNA, we treated cell lysates with Benzonase for 30 min at 4⁰C with end-to-end rotation. The WCE was centrifuged to pellet any cellular debris. GFPtagged viral proteins were immunoprecipitated with anti-GFP antibody (G10362, Life Technologies) overnight followed by a 2-hour incubation with Protein G Dynabeads (Invitrogen).

The beads were washed 3 times with buffer (10mM TRIS-HCl, pH7.9, 420mM NaCl, 0.1% NP-40) with end-to-end rotation in the cold room and twice with buffer without detergent (10mM TRIS-HCl, pH7.9, 420mM NaCl). The immunoprecipitated proteins were eluted with 0.5M NH4OH and lyophilized.

Each purified elute was digested in-solution with trypsin for MS analysis. Briefly, each sample was resuspended in 44uL of 50mM NH4HCO3, reduced with 100mM TCEP-HCL, alkylated with 500mM iodoacetamide for 45 min in the dark room, and digested with 1ug of trypsin overnight at 37°C. Samples were desalted using ZipTip Pipette tips (EMD Millipore) using standard procedures. The desalted samples were analyzed with an LTQ-Orbitrap Velos mass spectrometer (ThermoFisher Scientific) utilizing a 90-minute HPLC gradient and top 15 data-dependent acquisition.

We processed independently at least two biological replicates for each bait along with negative controls in each batch of sample. Material from HEK293 cells expressing GFP only was used as control. We performed extensive washes between samples to minimize carry-over. Furthermore, the order of sample acquisition on the mass spectrometer was reversed for the second replicate to avoid systematic bias. 

We used Cytoscape (V3.4.0; 47 ) to generate protein-protein interaction networks. For better illustration, individual nodes were manually arranged in physical complexes. Dot plots and heatmaps were generated using ProHits-viz 48 . Functional enrichment of PPI data was performed using ShinyGO (v0.61) 49 which utilizes hypergeometric distribution followed by FDR correction (FDR cutoff was 0.05). All MS files used in this study were deposited at MassIVE (http://massive.ucsd.edu).

The crystal structures for G3BP1 and the N protein were downloaded from the Protein Data Bank (PDB) to build their complex structure model (PDB ID: 4FCJ_A and 6M3M_B). The amino acid sequence for each protein was extracted from the corresponding structure files according to the 'ATOM' records. We generated MSA for each protein by running HHblits 50 with 8 iterations and E-value=1E-20 to search through the Uniclust30 library. The G3BP1 had 9435 homologous sequences and the N protein had 2665. These sequences were paired based on genomic distance or phylogeny. However, only the origin sequence was left after this concatenation. Thus, we input the concatenated single sequence to the deep learning-based algorithm trRosettato 51 to predict the interface residues. Based on the trRosetta prediction, we found the residues 90-96 (light blue spheres in Figure 1 ) of G3BP1 (blue cartoon) tend to interact with the residues 130-134 (orange spheres) of the N protein (orange cartoon). The above interface residues were used as distance constraints to build a complex structure model (Figure 1 ) with the HDOCK server 52 . Finally, we utilized the Rosetta docking protocol to optimize the interface with local refinement 53 . The interface energy of the final model was -5.46, which indicates a reliable model according to the Rosetta document 53 .

HEK293 cells were treated with 0.5mM sodium arsenite (NaAsO2) for one hour. Briefly, we used HEK293 cells expressing SARS-CoV-2 N protein tagged with GFP (GFP-N), GFP alone, and 'GFP-N + untagged G3BP1'. The expression was induced with doxycycline (1g/ml) for 24 hours prior to NaAsO2 treatment for one hour.

Immunofluorescence SARS-CoV-2 GFP-N and GFP expressing HEK293 cells were seeded on poly-L-lysine coated and acid-washed coverslips. The expression of the proteins was induced using 1g/ml doxycycline for 24 hours. Sodium arsenite (NaAsO2) treatment was performed as described above for 60 minutes prior to cell fixation. NaAsO2 was removed and cells were washed three times with PBS. Cells were fixed in 4% Paraformaldeyde for 15 minutes. Cells were subsequently permeabilized with 0.2% Triton X-100 in PBS for 5 min and incubated with block solution (1% goat serum, 1% BSA, 0.5% Tween-20 in PBS) for 1 hour. Santa Cruz G3BP (H-10) antibody was used for staining at 1:100 concentration in block solution for 2 hours at room temperature (RT). Cells were incubated with Goat anti-mouse secondary antibody and Hoescht stain in block solution for 1 hour at room temperature. Cells were fixed in Dako Fluorescence Mounting Medium (S3023). Imaging was performed the next day using a Zeiss confocal spinning disc AxioObserverZ1 microscope equipped with an Axiocam 506 camera using Zen software. A single focal plane was imaged, and stress granule quantification was performed for each replicate (n=2; number of cells 50). The number of stress granules per cell was plotted as box plot, and statistical significance was calculated using the student's t test.

Total RNA was extracted using the RNeasy extraction kit (Qiagen) following the manufacturer's instructions. Two independent biological samples for each condition were generated, resulting in a total of eight samples. DNase-treated total RNA was then quantified using Qubit RNA BR (cat # Q10211, Thermo Fisher Scientific Inc., Waltham, USA) fluorescent chemistry and 1 ng was used to obtain RNA Integrity Number (RIN) using the Bioanalyzer RNA 6000 Pico kit (cat # 5067-1513, Agilent Technologies Inc., Santa Clara, USA). Lowest RIN was 9.3; median RIN score was 9.75. 

Alignment and Read Processing iCLIP libraries were demultiplexed using XXXNNNNXX barcodes, where X is a random nucleotide. The 5'barcode and the illumine adaptor was trimmed using Cutadapt 54 (ver 2.10). The PCR duplicates were collapsed using UMI-tools 55 (ver 1.0.1). RNA-seq and iCLIP library reads were mapped to Gencode assembly 56 (GRCh37.p13) using STAR 57 (ver 2.7.1). Only the uniquely mapping reads were used for the downstream analyses.

To identify the differentially expressed genes from RNA-seq data, we used DESeq2 58 (ver 3.11) on gene counts generated using STAR and the human Gencode annotation V19. We filtered out genes with less than 10 counts across the sum of all RNA-seq samples. To plot differentially expressed genes as volcano plots, we used R-package EnhancedVolcanoplot (https://github.com/kevinblighe/EnhancedVolcano). The Gene Ontology enrichment analysis was performed and visualized using clusterProfiler 59 (ver 3.16.1), with the universe set to all the genes detected by RNA-seq. Multiple testing correction was performed using the Benjamini-Hochberg method and q-value cut off of 0.05 was used. Additionally, GO/ KEGG enrichment was performed using ShinyGO (v0.61), which utilizes hypergeometric distribution followed by FDR correction where FDR cutoff was set to 0.05.

Significant iCLIP peaks were called using Pureclip 60 , with input control and default settings, and the cross-linking sites within 50 bps from each other were merged. To visualize the distribution of N protein on genes, we used MetaPlotR 61 to calculate and scale the distance of N protein peaks relative to transcriptomic features. The processed PAR-CLIP peaks for G3BP1 and G3BP2 were downloaded from GEO (accession code: GSE98856) 32 . The read densities over ±1000 bp regions surrounding G3BP1 and G3BP2 peaks were calculated by diving the CLIP bpm over input bpm, where the bin size was set to 100. We used bedtools 62 (ver 2.29.2) to retrieve DNA sequences from the iCLIP peaks and subjected the sequences to motif analysis using the MEME-ChIP suite 63 (version 5.1.1). For nuclear counterstaining DAPI was used. 

A pneumonia outbreak associated with a new coronavirus of probable bat origin

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

Ribonucleocapsid Formation of Severe Acute Respiratory Syndrome Coronavirus through Molecular Action of the N-Terminal Domain of N Protein

Evaluation of nucleocapsid and spike protein-based enzyme-linked immunosorbent assays for detecting antibodies against SARS-CoV-2

The Coronavirus Nucleocapsid Is a Multifunctional Protein

The SARS coronavirus nucleocapsid protein -Forms and functions

Biochemical characterization of SARS-CoV-2 nucleocapsid protein

Crystal structure of SARS-CoV-2 nucleocapsid protein RNA binding domain reveals potential unique drug targeting sites

Architecture and self-assembly of the SARS-CoV-2 nucleocapsid protein

Nucleocapsid protein of SARS-CoV-2 phase separates into RNA-rich polymerase-containing condensates

SARS-CoV-2 nucleocapsid protein undergoes liquid-liquid phase separation stimulated by RNA and partitions into phases of human ribonucleoproteins

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

Virus-Host Interactome and Proteomic Survey Reveal Potential Virulence Factors Influencing SARS-CoV-2

Rasputin a decade on and more promiscuous than ever? A review of G3BPs

Regulation of stress granules in virus systems

Phosphorylation of G3BP1-S149 does not influence stress granule assembly

G3BP-Caprin1-USP10 complexes mediate stress granule condensation and associate with 40S subunits

The RasGAP-associated endoribonuclease G3BP assembles stress granules

G3BP1 Is a Tunable Switch that Triggers Phase Separation to Assemble Stress Granules

New World and Old World Alphaviruses Have Evolved to Exploit Different Components of Stress Granules, FXR and G3BP Proteins, for Assembly of Viral Replication Complexes

Zika Virus Hijacks Stress Granule Proteins and Modulates the Host Stress Response

A proposed role for the SARS-CoV-2 nucleocapsid protein in the formation and regulation of biomolecular condensates

Immunological characterizations of the nucleocapsid protein based SARS vaccine candidates

The Nucleocapsid Protein of SARS-CoV-2 Abolished Pluripotency in Human Induced Pluripotent Stem Cells

SAINTexpress: improvements and additional features in Significance Analysis of INTeractome software

Structural basis of RNA recognition by the SARS-CoV-2 nucleocapsid phosphoprotein

Crystal structures of the human G3BP1 NTF2-like domain visualize FxFG Nup repeat specificity

Ataxin-2-Like Is a Regulator of Stress Granules and Processing Bodies

N6-methyladenosine (m6A) recruits and repels proteins to regulate mRNA homeostasis

The RNA-Binding Protein Rasputin/G3BP Enhances the Stability and Translation of Its Target mRNAs

G3BP1 inhibits RNA virus replication by positively regulating RIG-Imediated cellular antiviral response

A direct RNA-protein interaction atlas of the SARS-CoV-2 RNA in infected human cells

Viral-mediated stabilization of AU-rich element containing mRNA contributes to cell transformation

The hepatitis C viral nonstructural protein 5A stabilizes growth-regulatory human transcripts

Conserved GU-Rich Elements Mediate mRNA Decay by Binding to CUG-Binding Protein 1

AU-rich elements: characterization and importance in mRNA degradation

Turnover of AU-richcontaining mRNAs during stress: A matter of survival

Endoplasmic reticulum as a potential therapeutic target for covid-19 infection management?

Why do SARS-CoV-2 NSPs rush to the ER

Human-chromatin-related protein interactions identify a demethylase complex required for chromosome segregation

A comprehensive, flexible collection of SARS-CoV-2 coding regions

iCLIP: protein-RNA interactions at nucleotide resolution

Multilayered Control of Alternative Splicing Regulatory Networks by Transcription Factors Resource Multilayered Control of Alternative Splicing Regulatory Networks by Transcription Factors

Multiparameter functional diversity of human C2H2 zinc finger proteins

The MaxQuant computational platform for mass spectrometry-based shotgun proteomics

Integration of biological networks and gene expression data using Cytoscape

ProHits-viz: A suite of web tools for visualizing interaction proteomics data

ShinyGO: A graphical gene-set enrichment tool for animals and plants

HHblits: Lightning-fast iterative protein sequence searching by HMM-HMM alignment

Improved protein structure prediction using predicted interresidue orientations

The HDOCK server for integrated protein-protein docking

Protein-protein docking with simultaneous optimization of rigid-body displacement and side-chain conformations

Cutadapt removes adapter sequences from high-throughput sequencing reads

UMI-tools: Modeling sequencing errors in Unique Molecular Identifiers to improve quantification accuracy

GENCODE reference annotation for the human and mouse genomes

STAR: Ultrafast universal RNA-seq aligner

Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2

ClusterProfiler: An R package for comparing biological themes among gene clusters

PureCLIP: capturing target-specific protein-RNA interaction footprints from single-nucleotide CLIP-seq data

A Perl/R pipeline for plotting metagenes of nucleotide modifications and other transcriptomic sites

BEDTools: a flexible suite of utilities for comparing genomic features

Authors would like to thank Dr. Frederick P. Roth for providing the SARS-