key: cord-0734082-rw94yls8
authors: Dominguez Andres, Ana; Feng, Yongmei; Campos, Alexandre Rosa; Yin, Jun; Yang, Chih-Cheng; James, Brian; Murad, Rabi; Kim, Hyungsoo; Deshpande, Aniruddha J.; Gordon, David E.; Krogan, Nevan; Pippa, Raffaella; Ronai, Ze’ev A.
title: SARS-CoV-2 ORF9c Is a Membrane-Associated Protein that Suppresses Antiviral Responses in Cells
date: 2020-08-19
journal: bioRxiv
DOI: 10.1101/2020.08.18.256776
sha: f0e803849f34a89afbf2909464ae44add28b99db
doc_id: 734082
cord_uid: rw94yls8

Disrupted antiviral immune responses are associated with severe COVID-19, the disease caused by SAR-CoV-2. Here, we show that the 73-amino-acid protein encoded by ORF9c of the viral genome contains a putative transmembrane domain, interacts with membrane proteins in multiple cellular compartments, and impairs antiviral processes in a lung epithelial cell line. Proteomic, interactome, and transcriptomic analyses, combined with bioinformatic analysis, revealed that expression of only this highly unstable small viral protein impaired interferon signaling, antigen presentation, and complement signaling, while inducing IL-6 signaling. Furthermore, we showed that interfering with ORF9c degradation by either proteasome inhibition or inhibition of the ATPase VCP blunted the effects of ORF9c. Our study indicated that ORF9c enables immune evasion and coordinates cellular changes essential for the SARS-CoV-2 life cycle. One-sentence summary SARS-CoV-2 ORF9c is the first human coronavirus protein localized to membrane, suppressing antiviral response, resembling full viral infection.

SARS-CoV-2 is an enveloped, positive-sense single strand 29.9 kb RNA virus (1, 2) that causes severe respiratory disease in humans . This coronavirus was first identified in Wuhan, China, at the end of 2019 (3) . Due to its easy human-to-human transmission and the lack of effective antiviral therapy, COVID-19 has caused a pandemic with more than 19 million cases and over 740,000 deaths worldwide (https://covid19.who.int). Mechanistically, the host protein ACE2 serves as the viral receptor and host cellular proteases, such as TMPRSS2, play key roles in SARS-CoV-2 entry into host cells (4) (5) (6) (7) . ACE2 expression is high in alveolar epithelial cells (8) , making the lung a highly vulnerable target for the virus.

SARS-CoV-2 infection causes a wide range of disease, from asymptomatic to mild disease to severe disease that can lead to death (9) . SARS-CoV-2 is most similar to the coronaviruses SARS-CoV and MERS-CoV (10, 11) . However, neither of those became a global pandemic.

Current therapies are primarily palliative and supportive (9) . More than 2000 clinical trials are currently in progress worldwide (12) (https://clinicaltrials.gov/ct2/who_table). Without effective vaccines or treatments, there is an urgent need to understand the pathology of SARS-CoV-2 infection, the roles of each of the 29 proteins encoded within the viral genome in the life cycle, virulence, and pathogenicity of the virus, and identify strategies for intervention or treatment.

Various therapeutic and vaccine strategies target viral entry mechanisms, such as vaccines or antibodies targeting on the Spike (S) protein (13) (14) (15) (16) ; others target viral replication or assembly processes, such as the antiviral drug remdesivir, which interferes with RNA replication and has emerged as superior to placebo in shortening recovery time in adults (17). Another strategy for treatment is interfere with viral immune evasion mechanisms and thus enable the body's natural 4 antiviral responses to be more effective at clearing the virus. Indeed, investigation of mechanisms of immune evasion by SARS-CoV-2 is an active area of translational research with immune evasion properties discovered for nonstructural protein 1 (Nsp1) (18) .

The SARS-CoV-2 genome contains 15 open reading frames (ORFs), which encode 29 viral proteins (19) (20) (21) . ORF1a and ORF1ab encode polyproteins that are cleaved into 16 nonstructural proteins (Nsp1 -Nsp16) that comprise the replicase-transcriptase complex. Spike (S) is encoded by ORF2, envelope (E) by ORF4, membrane (M) by ORF5, and nucleocapsid (N) by ORF9. An additional 9 ORFs encode "accessory" proteins: ORF3a, ORF3b, ORF6, ORF7a, ORF7b, ORF8, ORF9b, ORF9c, and ORF10.

Various studies have investigated the functions of the virally encoded proteins by performing interactome analysis in cells expressing individual viral proteins (19) or by evaluating the proteomic or transcriptomic changes associated with either viral infection (22) (23) (24) (25) (26) (27) . Others have used computational approaches to investigate protein-protein interactions between SARS-CoV-2 viral proteins and host proteins (28) . The interactome and proteome studies identified cellular processes affected by SARS-CoV-2 infection or specific viral proteins, notably innate immune signaling (19, 20, 23, (28) (29) (30) , ubiquitin ligase activities (19, 20, 23, (28) (29) (30) , p38 mitogenactivated protein kinase (MAPK) signaling (19, 20, 23, (28) (29) (30) . The transcriptomic studies identified interferon signaling (24) , cell death (27) , interleukin 1 (IL-1), IL-6, and chemokine signaling (22) .

Given the intense interest in catalytically active CoV-2 proteins (31) (32) (33) (34) , we examined the lessstudied group of ORFs encoding accessory proteins, which are largely thought to maintain viral 5 structural organization in replication organelles and within the viral particle (35, 36) . Here, we showed that expression of only ORF9c is sufficient to alter cellular networks in a manner that resembles full SARS-CoV-2 virus infection.

Seven of the 29 CoV-2 proteins are ORFs that lack catalytic activity and, in some cases, lack a known function (19) . Each of these were tagged with Strep at the N terminus and expressed them individually in the lung cancer epithelial cell line A549 in the presence or absence of the proteasome inhibitor MG132 (Fig. 1A ). The protein encoded by ORF9c was particularly unstable, with a profound increase in abundance evident in MG132-treated compared to that in vehicle-treated A549 cells (Fig 1A) .

ORF9c is present in previously characterized strains of SARS-CoV (37), a conservation suggesting a function in coronavirus pathogenesis. Phylogenetic analysis and alignment of the protein sequences showed that mutations are present in ORF9c among different coronavirus strains with bat SARS-like coronavirus ORF14 as the closest ortholog sharing 94% sequence identity and only 77% identity with ORF14 of SARS-CoV (Fig. 1B, fig. S1A ). TMHMM analysis (38) of SARS-CoV-2 ORF9c predicted a transmembrane sequence in the C-terminal domain, a motif not present in SARS-CoV-1 (or other human coronaviruses) ORF9c sequence ( Fig. 1B, fig. S1B ). Additionally, a single nucleotide mutation in SARS-CoV-2 ORF9c altered a termination codon, enabling the reading frame to extend by 3 amino acids (Fig. 1C ).

We assessed potential functions of ORF9c by performing transcriptome, interactome, and proteome analysis of A549 lung cancer cells transfected with ORF9c tagged at the N terminus with 2 copies of the Strep tag (19) . To map the ORF9c interactome, we conducted liquid chromatography tandem mass spectrometry (LC-MS/MS) of 2xStrep-tagged ORF9c compared with control 2xStrep-tagged GFP) immunoprecipitated from A549 cells 24 h after transfection.

ORF9c interactome analysis revealed that most interacting proteins were classified as membrane proteins (Fig. 1D , table S1) according to Gene Ontology Cellular Component. As a protein with a transmembrane domain, this was not surprising. However, we were surprised to find that the ORF9c interactome was distributed throughout the membrane-bound organelles (Fig. 1D ), including >30 proteins in the protein biosynthesis and transport systems of the endoplasmic reticulum (ER) and Golgi, >15 proteins in the mitochondria, and >30 other membrane-related proteins. Given the instability of ORF9c, we were not surprised to identify a group of membraneassociated proteins that function with the proteasome. Comparison between ORF9c and ORF10, both expressed using the Strep-tagged vector, confirmed that enrichment of membranal proteins as part of the interactome was selectively seen for the ORF9c (fig. S1C).

We conducted both label free quantification (LFQ) and tandem mass tag (TMT) mass spectrometry analysis to identify changes in the cellular proteome in A549 cells expressing ORF9c. We compared the proteomic changes associated with ORF9c expression in the presence 7 or absence of proteasomal inhibition with MG132, using DMSO as the vehicle control for comparison in each set. Principle component analysis (PCA) revealed that ORF9c contributed to major variance in all data sets ( fig. S2A, S2B ). Pairwise comparisons between ORF9c and control untransfected samples identified differentially expressed proteins in both the DMSO and MG132 groups (Fig 2A) . In both the DMSO and MG132 datasets, most changes induced by ORF9c were a reduction in protein abundance (downregulation) ( Fig. 2A, table S1 ).

Downregulated proteins identified using both approaches consistently showed ~60% overlap, while no overlap were identified among the upregulated proteins. Thus, to maximize the discovery of ORF9c dysregulated proteins, results from both technologies were combined.

Including the differentially regulated proteins identified by both the TMT and LFQ analysis revealed 14 proteins were upregulated in common by ORF9c expression in the presence or absence of the proteasome inhibitor and 144 proteins were downregulated in common (Fig. 2B ).

Using the downregulated proteins and upregulated proteins identified for either the DMSO or MG132 condition separately, we performed Ingenuity Pathway Analysis (IPA) to assess signaling pathways deregulated in ORF9c-expressing cells. In both the DMSO and MG132 condition, interferon (IFN) signaling exhibited the greatest difference, both in terms of the intensity of the downregulation and the number of proteins significantly associated with this pathway, in response to ORF9c (Fig. 2C, table S1 ). Other pathways affected by ORF9c and of particular importance to virulence were antigen presentation and innate immune response pathways, such as IRF/cytosolic pattern recognition receptors. We further examined potential upstream regulators of these pathways using Ingenuity Pathway Analysis for proteins that exhibited a change in abundance in the ORF9c-expressing cells. This analysis revealed that several components of the IFN machinery [interferons (IFNL1, IFNA), interferon responsive 8 transcription factors (IRF7, IRF1), and an interferon receptor (INFAR)] were reduced, consistent with the impaired IFN signaling, and an increase in MAPK1 (also known as ERK2) abundance (Fig. 2C, table S1 ).

To assess if there were notable differences in the intensity of the changes in protein abundance in response to proteasome inhibition, we calculated relative changes in protein abundance between control and ORF9c-expressing cells from both the DMSO and MG132 conditions for proteins associated with IFN signaling or the ubiquitin proteasome (UBP) system and antigen presentation (Fig. 2D ). The intensity of the changes was similar in the presence or absence of MG132, suggesting even small amounts of the unstable ORF9c are sufficient to induce cellular changes including those that contribute to immune evasion.

Consistent with the IPA-based analysis (Fig. 2B) , IFN signaling components, including IFI35, multiple IFIT proteins, IRF9, ISG15, MX1, PSMB8, and STAT proteins, were downregulated in ORF9c-expressing cells in both the presence and absence of MG132 (Fig. 2D, left) . Indicative of a decrease in antigen presentation capacity, multiple proteins involved in this process were decreased, including proteins involved in antigen loading and display [HLA proteins, β2M, and antigen transporters (TAP1 and TAP2)] and proteins involved in UBP [ubiquitin-conjugating enzymes UBE2I and UBE2L6), deubiquitination enzymes (USP18 and UPS41), and proteasome components (PSMB and PSME proteins)] (Fig 2D, right) .

These changes in the proteome indicated that the expression of only ORF9c, even in the absence of proteasomal inhibition to stabilize this protein, is sufficient to elicit effective inhibition of IFN, immune recognition, and UBP components at the protein level. Such a response suggested that ORF9c contributes to immune evasion of SARS-CoV-2.

We assessed transcriptional changes elicited by ORF9c expression in A549 cells using RNA-seq analysis. PCA showed that changes induced by ORF9c in both DMSO-and MG132-treated cells cluster in distinct experimental groups (Fig. S2C) . In contrast to the proteomic results that revealed predominant downregulation of proteins following ORF9c expression, RNA-seq analysis showed a similar number of transcripts were increased or decreased in the presence or absence of MG132 (Fig. 3A, table S2 ). Additionally, the number of differentially regulated transcripts was higher than that for the differentially regulated proteins.

Using IPA, we identified the pathways significantly enriched in differentially regulated transcripts. The same set of pathways were identified in the DMSO and MG132 conditions, and similar to the proteomic results, most related to immune signaling (Fig. 3B , upper, table S2).

However, many of the specific pathways were different from those identified at proteomic level.

At the transcriptional level, we detected the greatest effects on the complement system and several pathways involved in inflammatory signaling. Thus, some components of antigen presentation and immune signaling pathways showed comparable changes at the protein and mRNA levels; other changes elicited by ORF9c expression were unique to the transcriptional level, such as induction of IL-6 signaling and p38 MAPK signaling, or the protein level, such as impairment of IFN signaling.

We analyzed the ORF9c-regulated transcripts for those encoding upstream regulators of the pathways altered at the transcriptional level by ORF9c. This analysis identified the classic immune modulators tumor necrosis factor (TNF), IL-1B, IFNg, transforming growth factor β (TGFb), and NF-kB signaling components (Fig. 3B , lower, table S2).

We evaluated transcripts associated with the complement system or IL-6 signaling in detail. In both the presence and absence of proteasome inhibition, complement system transcripts were mostly downregulated by ORF9c expression in A549 cells (Fig. 3C, left) . For IL-6 signaling, we found some differences between the MG132 and DMSO conditions (Fig. 3C, right) . For several transcripts the intensity of the upregulation was greater in the absence of proteasome inhibition (MAP3K14 and MAP2K3, IL6 and IL6R, SOCS1 and SOCS3); for others the presence of the proteasome inhibitor resulted in a greater reduction in transcript abundance (IL1B, TNFAIP6, CD14, IL1R2). Thus, these results suggested that ORF9c had a dose-dependent effect on some transcripts.

We combined the results of the proteomic and transcriptomic (Fig. 4A ), which revealed a small set of commonly upregulated or downregulated genes by ORF9c at both the transcription and protein levels (Fig. 4B ). We performed IPA canonical pathway analysis and found commonly altered pathways at both the transcript and protein levels (Fig. 4C , table S1, S2). The direction of regulation (increased or decreased activity) was consistent between the transcripts and proteins.

However, the number of components significantly enriched in most of the pathways differed between the protein and transcript levels.

We compared our findings at the transcriptome, proteome, and interactome levels with those reported by Stukalov et al. (39) for proteins with altered ubiquitination (ubiquitinome) in response to SARS-CoV-2 infection of A549 cells. Within the top 10 enriched IPA canonical pathways, we noticed enrichment across all 5 protein ubiquitination data sets, sirtuin signaling, phagosome maturation, tight junction signaling, and caveolar-mediate endocytosis (Fig. 4D , table S3). Seven of the 10 were common between the ubiquitinome data (39) and our proteomic analyses by LFQ or TMT mass spectrometry.

We also compared the pathway enrichment across our transcriptomic data and those from Blanco-Melo et al. (22) reporting transcriptomic changes upon SARS-CoV-2 infection in human primary epithelial cells or ACE-expressing A549 cells and with those from Stukalov et al. (39) reporting transcriptomic changes in ACE2-expressing A549 cells 12 and 24 hours after infection 

In the presence or absence of the proteasome inhibitor, we observed substantial overlap (55 -65%) in the ORF9c-downregulated proteins ( Fig. 2A) and 54 -72% overlap in ORF9cupregulated transcripts and 45 -64% in the downregulated transcripts (Fig. 3A) . Thus, despite

ORF9c increasing in MG132-treated A459 cells, the persistent changes in MG132-treated cells suggested that even a low amount of ORF9c is sufficient to elicit its cellular effects. However, some transcripts (399) and proteins (97) were downregulated by ORF9c in cells not treated with MG132 and were upregulated in cells treated with MG132 ( Fig. 5A ).

Pathway analysis on the transcripts and proteins showed discordant regulation in DMSO-or MG132-treated cells in response ORF9c. The most pronounced among all three datasets were components of the UBP and the unfolded protein response (UPR) (Fig. 5B, table S4 ). Changes in events or pathways associated with the cell cycle was not surprising given the critical role UBP plays in their regulation. We thus hypothesized that both the UBP and UPR were involved in degradation of ORF9c. Finding that UPR signaling was also reversed upon MG132 treatment ( Fig 5B) is consistent with the role of UPR signaling in degradation of ORF9c.

To directly assess the importance of UBP and UPR components in ORF9c instability, we performed an siRNA-based screen targeting over 1100 genes that encode components of both machineries in A549 cells stably expressing Strep-tagged ORF9c (Fig. S3 ). The top hits independently validated as blocking ORF9c degradation were siRNAs targeting VCP [also known as p97, an ATPase involved in export of unfolded proteins from the ER for ER-associated degradation (ERAD) and in ER to Golgi transport] (40, 41) , the proteasomal subunit PSMD2, and the proteasome maturation factor POMP (42) , which has been also implicated in ERAD (43) 13 and in IFN-induced reorganization of proteasomes into immunoproteasomes (44) (Fig. 5C , Table   1 ).

To assess if interfering with ERAD affected the cellular effects induced by ORF9c, we compared the transcript abundance for 6 genes (IFNGR1, IGS15, IRF9, SOCS1, PSMB8, TAP1) that were downregulated by expression of ORF9c in DMSO-treated cells with their abundance in cells treated with the VCP inhibitor MNS-873, the heat shock protein 90 (HSP90) inhibitor geldanamycin, or the proteasome inhibitor bortezomib. Although the HSP90 inhibitor and the proteasome inhibitor increased transcript abundance for some of the gens tested, VCP inhibition was the most consistently effective at enabling expression of each of these transcripts in the ORF9c-expressing cells (Fig. 5D ). These observations suggested that ORF9c ability to attenuate key cellular signaling involved in antiviral responses, including antigen presentation, immune, and IFN pathways, requires the activity of VCP.

The key to our ability to control spread of the SARS-CoV-2 virus is to understand its mechanism of action and how the concerted action of its 29 encoded proteins subvert cellular regulatory networks. One can divide viral "success" into two key phases: infection, which is the ability to enter a given cell type, and multiplication that enables continuous infection through viral replication and packaging, which exploits host cell machineries (45) . A third aspect to viral success is evasion from immune clearance. Disruption of either the infection or replication phases should effectively inhibit the SARS-CoV-2 life cycle. Accordingly, many efforts focus on neutralizing interaction of the viral S protein with ACE2 (4, 46) . Other efforts strive to interfere 14 with the viral life cycle after it invades target cells, and many focus on catalytically active proteins encoded by the SARS-CoV-2 genome (47).

Here, we analyzed one small, unstable SARS-CoV-2 protein, ORF9c in the context of an epithelial lung cancer cell line. Limited overlap between published studies of the ORF9c interactome can be attributed to their use of different cell system (HEK293 compared with A549 lung cancer cells used here, as the use of different filtering criteria (19) . However, many of the cellular changes elicited by ORF9c in our study also occur following infection with full replicative SARS-CoV-2 virus (22) . Those phenotypes included changes in IFN and other cytokine signaling, immune recognition (including antigen presentation; dendritic cell, T cell, and acute immune responses; and pattern recognition), cell cycle, and the complement system, all of which were downregulated by ORF9c. Additionally, similar to cells infected with the virus or expressing ORF9c, IL-1, IL-6, and p38 MAPK signaling pathways were upregulated. The primary change identified in our analysis was deregulation of the IFN system, coupled with changes in cytokines associated with TNF and STAT signaling and factors implicated in innate immunity. In addition to mediating an antiviral response, aberrant IFN signaling is also critical for numerous pathological indications linked to COVID-19 (48). Thus, we concluded that SARS-CoV-2 ORF9c elicits pathologies not seen with previously characterized coronavirus prototypes, primarily through effective modulation of IFN signaling. Our findings suggested that ORF9c enables cells to escape from immune surveillance through by reducing HLA abundance and antigen presentation, while also slowing cell replication, which could viral replication of infected cells.

Strikingly, ORF9c is predicted have a transmembrane domain and we found that the ORF9c interactome was mostly comprised of membrane-associated proteins in multiple organelles, 15 including ER, Golgi, mitochondria, cell surface membrane, and peroxisomes. Indeed, many of the cellular changes that we observed following ORF9c expression are associated with membrane proteins or pathways mediated by proteins that associate with the membranes of various cellular compartments. Importantly, SARS-CoV-2 ORF9c is the first human coronavirus ORF9c protein that has acquired this putative transmembrane sequence. Mutations have been acquired along the course of evolution of ORF9c, although ~80% of the SARS-CoV-2 ORF9c sequence is identical to the ortholog in other coronaviruses, although greater similarity was identified with the bat SARS-CoV-2 sequence. The membranal anchoring capability identified in SARS-CoV-2 ORF9c is novel feature that may mediate the effect on IFN signaling, antigen presentation, and immune evasion phenotypes, characteristics that make SARS-CoV-2 much more virulent and pathogenic than other coronaviruses. Notably, 0.7-1.4% of patients were found to possess a mutation that is expected to impair transmembrane domain of SARS-CoV-2 ORF9c (19) ; awaiting future assessment of clinical outcome, our data would predict a better clinical outcome, distinguishing these patients from those harboring the transmembrane domain.

Correspondingly, the interactome for the less virulence and pathogenic SARS-CoV-1 ORF9c (49) did not overlap with that for SARS-CoV-2 ORF9c.

A notable signature that we identified is the upregulation of histone and histone deacetylaserelated factors, which suggested that histone modification may underlie the transcriptional repression. The increased transcription of AP1 family members (FOSB, FOSL1, CREB, and ATF3), which participate in the cellular stress response, may reflect the response to stress imposed by ORF9c, which, in turn, can limit immune-related signaling identified in our study.

Another remarkable signature of ORF9c expression in A459 cells was the association with UBP components. Together with the observations on cellular immune pathways, this association with 16 the UBP suggested that ORF9c induces changes in UBP components that alter the stability of cellular proteins implicated in cytokine signaling, antigen presentation, innate immunity, and the cell cycle. Additionally, we identified UPR components as important for ORF9c instability, suggesting that this protein is misfolded or at least recognized as a misfolded protein by the host cell. In this scenario, we propose that misfolded ORF9c engages UPR (through VCP) and the UBP, which clears this protein. By engaging the UBP, ORF9c promotes enhanced proteasome activity as suggested by our proteome analysis.

Our analysis revealed that interfering with VCP activity blunted the transcriptional repressive effects of SARS-CoV-2 ORF9c on impact immune system components, such as IRF9, INFGR1, ISG15, SOCS1 and TAP1. Proteasome inhibition with bortezomib was also effective, although not as consistently effective as MNS-873, the VCP inhibitor. These findings suggested that inhibition of VCP or the proteasome, which has inhibitors currently in clinical trials for cancer (50, 51), may be considered among therapeutic measures to fight SARS-CoV-2 virulence and pathologies. However, we cannot exclude the possibility that ORF9c stability and degradation mechanisms differ based on cell type or the activity of other viral proteins in infected cells.

Another potential therapeutic opportunity involves targeting the membrane association of ORF9c, because this is a unique feature of the protein in the SARS-CoV-2 coronavirus. Thus, identifying small molecules that could interfere with ORF9c localization to the membrane could limit ORF9c function and impede the ability of the virus to evade the immune response and reduce viral replication.

Given that ORF9c is expected to affect immune evasion, virulence and pathogenesis, additional studies should assess the consequences of ORF9c inhibition in vivo, using primates and possibly mouse models where SARS-CoV-2 shown to impact IFN signaling and immune response (48). Secondary antibodies were used at 1:5000.

Immunoprecipitation of streptavidin-tagged CoV-2 ORFs was performed as previously described (19) . Briefly, frozen cell pellets were thawed on ice for 15-20 minutes and suspended in 1ml

Lysis Buffer with 50 mM Tris-HCl, pH 7.4 at 4°C, 150 mM NaCl, 1 mM EDTA and supplemented with 0.5% Nonidet P-40 Substitute, Complete mini EDTA-free protease and PhosSTOP phosphatase inhibitor cocktails (Roche). Samples were centrifuged 10 minutes at 4°C at 13,000g. Protein quantification was performed using Pierce's BCA quantification kit as per the manufacturer's indications. Supernatants (1 mg protein) were incubated 2h at 4°C with MagStrep "type3" beads (30 μl; IBA Lifesciences) that had been previously equilibrated twice with 1 ml Wash Buffer (IP Buffer supplemented with 0.05% NP40). Beads were washed five times with 1 ml Wash Buffer and then five times with 1 ml Ammonium Bicarbonate 50mM.

Raw FASTQ files were processed using cutadapt v1.18 (52) Figure 4D , Figure 5A and 5B were selected based on BH correct P <0.05 without fold change cutoff. Differentially expressed proteins in Figure 4D , Figure 5A and 5B were selected based on P <0.01 without fold change cutoff.

Protein sequences similar to SARS-CoV-2 ORF9c were retrieved through NCBI Blastp using the nr database (58). Sequence alignment was performed using Clustalo (59). Phylogenetic tree was built using PhyML algorithm by 100 times bootstrap, and visualized using Seaview (60) and

Geneious version 2020.2.2 (San Diego, CA). Transmembrane domain prediction was performed using TMHMM web server v2.0 (38) . Transmembrane domain was predicted based on TMHMM posterior probability more than 0.5.

Statistical test results of RNA-Seq, proteomics and interactome data provided in Supplementary Tables. Analyses of omics data in this study were performed using R customized scripts.

Statistical analysis of proteomics data sets was performed using MSstats (label-free data) and

MStatsTMT (TMT data) bioconductor package. Differential expression of RNA-Seq was performed using DESeq2 bioconductor package following Negative Binomial Distribution and Wald test. Pathway enrichment and upstream regulator analyses were performed using IPA following Fisher's Exact Test and Z-score calculation considering directional changes in IPA database. 

Ana Dominguez Andres 1^, Yongmei Feng 1^, Alexandre Rosa Campos 1^, Jun Yin 1^, Chih-Cheng 

Beads were resuspended in 8M urea, 50 mM ammonium bicarbonate, and cysteine disulfide bonds were reduced with 10 mM tris (2-carboxyethyl) phosphine (TCEP) at 30°C for 60 min. High confidence interacting proteins were selected using the following filtering criteria: log2FC > 3.3 (10x) and a p-value <0.025 (to include the p-value of proteins detected in at least 2 ORF9c pulldown replicates but not detected in the negative controls). We also considered the 'crapomeScore' < 0.5, which is the fraction of single affinity purification experiments a given protein-interacting candidate receives in the Crapome database (crapome.org). A score of 1 means the candidate is identified in all experiments in that database.

Cells were lysed in UAB buffer (8M urea, 50 mM ammonium bicarbonate (ABC) and

Benzonase 24U/100ml) with vigorous shaking (20 Hz for 10 min at room temperature using a Retsch MM301 instrument). Lysates were centrifuged at 14,000xg for 10 minutes to remove cellular debris, and protein concentration in supernatants was determined using bicinchoninic acid (BCA) protein assay (Thermo Scientific Fragmented precursors were detected in the ion trap as rapid scan mode with automatic gain control target set to 1 x 10 4 and a maximum injection time set at 35 ms. The dynamic exclusion was set to 20 seconds with a 10 ppm mass tolerance around the precursor.

For processing label-free LC-MS/MS data, all raw files were processed with MaxQuant (version 1.5.5.1) using the integrated Andromeda Search engine against a target/decoy version of the curated human Uniprot proteome without isoforms (downloaded in January of 2020) and the GPM cRAP sequences (commonly known protein contaminants). First search peptide tolerance was set to 20 ppm, and main search peptide tolerance was set to 4.5 ppm. Fragment mass tolerance was set to 20 ppm. Trypsin was set as the enzyme in specific mode, and up to two missed cleavages was allowed. Carbamidomethylation of cysteine was specified as fixed modification and protein N-terminal acetylation and oxidation of methionine were considered variable modifications. In addition, the phosphopeptide-enriched samples were also searched with phosphorylation of serine, threonine or tyrosine considered as variable modification. The target-decoy-based false discovery rate (FDR) filter for spectrum and protein identification was set to 1%.

Statistical analysis of label-free proteomics data was carried out using in-house R script (version 3.5.1, 64-bit), including R Bioconductor packages. First, peptide feature intensities (MaxQuant   evidence table) were log2-transformed and normalized (loess normalization) across samples to account for systematic errors. Then all non-razor peptide sequences were removed from the list.

Protein-level quantification and statistical testing for differential abundance were performed using MSstats bioconductor package. 

Cells were imaged with an IC200 high-content screening system (Vala Sciences) using a 20X objective to visualize Strep-ORF9c proteins (Alexa 488) and nuclei (DAPI). Four images were obtained from different fields in each well for 384-well plates. Images were analyzed with Acapella high-content imaging and analysis software for valid cell numbers per field and to determine average Alexa 488 intensity per cell. A549 cells expressing SARS-CoV-2 Strep-ORF9c were treated with DMSO or MG132 (10uM) served as negative and positive imaging controls, respectively. Plate-to-plate variability was normalized using a control-based method; associated control samples were aggregated, and the mean and variance across wells were determined. The Alexa488 mean intensity for all wells with siRNA knockdown was normalized using unique non-targeting siRNAs included in each plate as reference data points. The top 36 scoring hits were obtained using a threshold of > 1.46-fold increase in average intensity from duplicates (p-value <0.05). Ten of the 36 siRNA pools were selected for confirmation in a secondary deconvolution screen. For that screen, quantification data were converted to a Z-score, and the average Z-score from data in triplicate plates was determined. Genes were defined as confirmed screen hits if they had 3 or more individual positive siRNA score (cut-off of >3 SD). Table S1 . Interactome and proteome data analysis, pathway enrichment and upstream regulators analyses Table S2 . Transcriptome data analysis, pathway enrichment and upstream regulators analyses Table S3 . Canonical pathway comparison of different omics technologies and public data sets Table S4 . Canonical pathway analysis of MG132 reversed genes and proteins

A novel coronavirus outbreak of global health concern

A new coronavirus associated with human respiratory disease in China

A Novel Coronavirus from Patients with Pneumonia in China

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

Cell entry mechanisms of SARS-CoV-2

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

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

Single-cell RNA Expression Profiling of ACE2, The Receptor of SARS-CoV-2

SARS-CoV-2: A comprehensive review from pathogenicity of the virus to clinical consequences

Genome Composition and Divergence of the Novel Coronavirus (2019-nCoV) Originating in China

Systematic Comparison of Two Animal-to-Human Transmitted Human Coronaviruses: SARS-CoV-2 and SARS-CoV

Characteristics of registered studies for Coronavirus disease 2019 (COVID-19): A systematic review

An Alphavirus-derived replicon RNA vaccine induces SARS-CoV-2 neutralizing antibody and T cell responses in mice and nonhuman primates

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

A vaccine targeting the RBD of the S protein of SARS-CoV-2 induces protective immunity

Potently neutralizing and protective human antibodies against SARS-CoV-2

Remdesivir for the Treatment of Covid-19 -Preliminary Report

Structural basis for translational shutdown and immune evasion by the Nsp1 protein of SARS-CoV-2

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

Structural Genomics of SARS-CoV-2 Indicates Evolutionary Conserved Functional Regions of Viral Proteins

The Proteins of Severe Acute Respiratory Syndrome Coronavirus-2 (SARS CoV-2 or n-COV19), the Cause of COVID-19

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

Proteomics of SARS-CoV-2-infected host cells reveals therapy targets

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

Transcriptional landscape of SARS-CoV-2 infection dismantles pathogenic pathways activated by the virus, proposes unique sex-specific differences and predicts tailored therapeutic strategies

A Dynamic Immune Response Shapes COVID-19 Progression

Transcriptomic characteristics of bronchoalveolar lavage fluid and peripheral blood mononuclear cells in COVID-19 patients

Interactome of SARS-CoV-2 / nCoV19 modulated host proteins with computationally predicted PPIs

The Global Phosphorylation Landscape of SARS-CoV-2 Infection

COVID-19: viral-host interactome analyzed by network based-approach model to study pathogenesis of SARS-CoV-2 infection

Drug design targeting the main protease, the Achilles' heel of coronaviruses

Crystal structure of SARS-CoV-2 main protease provides a basis for design of improved alpha-ketoamide inhibitors

Structure of the RNA-dependent RNA polymerase from COVID-19 virus

Structural and Evolutionary Analysis Indicate That the SARS-CoV-2 Mpro Is a Challenging Target for Small-Molecule Inhibitor Design

Characterization of accessory genes in coronavirus genomes. bioRxiv

SARS coronavirus accessory proteins

Insights into SARS-CoV-2 genome, structure, evolution, pathogenesis and therapies: Structural genomics approach

Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes

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

Ubiquilin and p97/VCP bind erasin, forming a complex involved in ERAD

VCP/p97-Mediated Unfolding as a Principle in Protein Homeostasis and Signaling

The proteasome maturation protein POMP facilitates major steps of 20S proteasome formation at the endoplasmic reticulum

siRNA silencing of proteasome maturation protein (POMP) activates the unfolded protein response and constitutes a model for KLICK genodermatosis

IFN-gamma-induced immune adaptation of the proteasome system is an accelerated and transient response

COVID-19 infection: Origin, transmission, and characteristics of human coronaviruses

SARS-CoV-2 Receptor ACE2 Is an Interferon-Stimulated Gene in Human Airway Epithelial Cells and Is Detected in Specific Cell Subsets across Tissues

Competing Interests: ZAR is a co-founder and serves as scientific advisor to Pangea Therapeutics. All other authors declare no competing interests.Data and materials availability: All datasets will be deposited in publicly available data sets prior to publication; all reagents and study protocols are available by requests from the corresponding authors.