key: cord-342189-ya05m58o
authors: Banerjee, Abhik K.; Blanco, Mario R.; Bruce, Emily A.; Honson, Drew D.; Chen, Linlin M.; Chow, Amy; Bhat, Prashant; Ollikainen, Noah; Quinodoz, Sofia A.; Loney, Colin; Thai, Jasmine; Miller, Zachary D.; Lin, Aaron E.; Schmidt, Madaline M.; Stewart, Douglas G.; Goldfarb, Daniel; De Lorenzo, Giuditta; Rihn, Suzannah J.; Voorhees, Rebecca; Botten, Jason W.; Majumdar, Devdoot; Guttman, Mitchell
title: SARS-CoV-2 disrupts splicing, translation, and protein trafficking to suppress host defenses
date: 2020-10-08
journal: Cell
DOI: 10.1016/j.cell.2020.10.004
sha: 
doc_id: 342189
cord_uid: ya05m58o

SARS-CoV-2 is a recently identified coronavirus that causes the respiratory disease known as COVID-19. Despite the urgent need, we still do not fully understand the molecular basis of SARS-CoV-2 pathogenesis. Here, we comprehensively define the interactions between SARS-CoV-2 proteins and human RNAs. NSP16 binds to the mRNA recognition domains of the U1 and U2 splicing RNAs and acts to suppress global mRNA splicing upon SARS-CoV-2 infection. NSP1 binds to 18S ribosomal RNA in the mRNA entry channel of the ribosome and leads to global inhibition of mRNA translation upon infection. Finally, NSP8 and NSP9 bind to the 7SL RNA in the Signal Recognition Particle and interfere with protein trafficking to the cell membrane upon infection. Disruption of each of these essential cellular functions acts to suppress the interferon response to viral infection. Our results uncover a multipronged strategy utilized by SARS-CoV-2 to antagonize essential cellular processes to suppress host defenses.

Coronaviruses are a family of viruses with notably large single-stranded RNA genomes and broad species tropism among mammals (Graham and Baric, 2010) . Recently, a coronavirus, SARS-CoV-2, was discovered to cause the severe respiratory disease known as COVID-19. It is highly transmissible within human populations and its spread has resulted in a global pandemic with more than a million deaths to date (Andersen et al., 2020; Zou et al., 2020) . We do not fully understand the molecular basis of infection and pathogenesis of this virus in human cells.

Accordingly, there is an urgent need to understand these mechanisms to guide the development of therapeutics.

SARS-CoV-2 encodes 27 proteins with diverse functional roles in viral replication and packaging( Bar-On et al., 2020; Wang et al., 2020) . These include 4 structural proteins: the nucleocapsid (N, which binds the viral RNA), and the envelope (E), membrane (M), and spike (S) proteins, which are integral membrane proteins. In addition, there are 16 non-structural proteins (NSP1-16) which encode the RNA-directed RNA polymerase, helicase, and other components required for viral replication (da Silva et al., 2020) . Finally, there are 7 accessory proteins (ORF3a-8) whose function in viral replication or packaging remain largely uncharacterized (Chen and Zhong, 2020; Finkel et al., 2020) .

As obligate intracellular parasites, viruses require host cell components to translate and transport their proteins and to assemble and secrete viral particles (Maier et al., 2016) . Upon viral infection, the mammalian innate immune system acts to rapidly detect and block viral infection at all stages of the viral life cycle (Chow et al., 2018; Jensen and Thomsen, 2012; Wilkins and Gale, 2010) .

The primary form of intracellular viral surveillance engages the interferon pathway, which amplifies signals resulting from detection of intracellular viral components to induce a systemic type I interferon response upon infection (Stetson and Medzhitov, 2006) . Specifically, cells contain various RNA sensors (such as RIG-I and MDA5) that detect the presence of viral RNAs, promote nuclear translocation of the transcription factor IRF3 leading to transcription, translation, and secretion of interferon (e.g. IFN-α and IFN-β) . Binding of interferon to cognate cell-surface receptors leads to transcription and translation of hundreds of antiviral genes.

J o u r n a l P r e -p r o o f 4 In order to successfully replicate, viruses employ a range of strategies to counter host antiviral responses (Beachboard and Horner, 2016) . In addition to their essential roles in the viral life cycle, many viral proteins also antagonize core cellular functions in human cells to evade host immune responses. For example, human cytomegalovirus (HCMV) encodes proteins that inhibit class 1 Major Histocompatibility (MHC) display on the cell surface by retaining MHC proteins in the endoplasmic reticulum (Miller et al., 1998) , polioviruses encode proteins that degrade translation initiation factors (eIF4G) to prevent translation of 5'-capped host mRNAs (Kempf and Barton, 2008; Lloyd, 2006) , and influenza A encodes a protein that modulates mRNA splicing to degrade the mRNA that encodes RIG-I (Kochs et al., 2007; Zhang et al., 2018) .

Suppression of the interferon response has recently emerged as a major clinical determinant of COVID-19 severity , with almost complete loss of secreted IFN characterizing the most severe cases (Hadjadj et al., 2020) . The extent to which SARS-CoV-2 suppresses the interferon response is a key characteristic that distinguishes COVID-19 from SARS and MERS (Lokugamage et al., 2020) . Several strategies have been proposed for how the related SARS-and MERS-causing viruses may hijack host cell machinery and evade immune detection, including repression of host mRNA transcription in the nucleus (Canton et al., 2018) , degradation of host mRNA in the nucleus and cytoplasm (Kamitani et al., 2009; , and inhibition of host translation (Nakagawa et al., 2018) . Nonetheless, the extent to which SARS-CoV-2 uses these or other strategies, and how they may be executed at a molecular level remains unclear.

Understanding the interactions between viral proteins and components of human cells is essential for elucidating their pathogenic mechanisms and for development of effective therapeutics.

Because SARS-CoV-2 is an RNA virus and many of its encoded proteins are known to bind RNA (Sola et al., 2011) , we reasoned that these viral proteins may interact with specific human mRNAs (critical intermediates in protein production) or non-coding RNAs (critical structural components of diverse cellular machines) to promote viral propagation.

Here, we comprehensively define the interactions between each SARS-CoV-2 protein and human RNAs. We show that 10 viral proteins form highly specific interactions with mRNAs or ncRNAs, including those involved in progressive steps of host cell protein production. We show J o u r n a l P r e -p r o o f 5 that NSP16 binds to the mRNA recognition domains of the U1 and U2 RNA components of the spliceosome and acts to suppress global mRNA splicing in SARS-CoV-2-infected human cells.

We find that NSP1 binds to a precise region on the 18S ribosomal RNA that resides in the mRNA entry channel of the initiating 40S ribosome. This interaction leads to global inhibition of mRNA translation upon SARS-CoV-2 infection of human cells. Finally, we find that NSP8 and NSP9 bind to discrete regions on the 7SL RNA component of the Signal Recognition Particle (SRP) and interfere with protein trafficking to the cell membrane upon infection. We show that disruption of each of these essential cellular functions acts to suppress the type I interferon response to viral infection. Together, our results uncover a multipronged strategy utilized by SARS-CoV-2 to antagonize essential cellular processes and robustly suppress host immune defenses.

We cloned all 27 of the known SARS-CoV-2 viral proteins into mammalian expression vectors containing an N-terminal HaloTag (Los et al., 2008) (Figure S1A , Methods), expressed each in HEK293T cells, and exposed them to UV light to covalently crosslink proteins to their bound RNAs. We then lysed the cells and purified each viral protein using stringent, denaturing conditions to disrupt any non-covalent associations and capture those with a UV-mediated interaction ( Figure 1A , Methods). As positive and negative controls, we purified a known human RNA binding protein (PTBP1) and a metabolic protein (GAPDH) (Figure S1A-E).

We successfully purified 26 of the 27 viral proteins (Figure S1A ; full-length Spike was not soluble when expressed). We found that 10 viral proteins (NSP1, NSP4, NSP8, NSP9, NSP12, NSP15, NSP16, ORF3b, N, and E protein) bind to specific host RNAs (p-value < 0.001, Figure   1B , Table S1), including 6 structural ncRNAs and 142 mRNAs (Table S1). These include mRNAs involved in protein translation (e.g. COPS5, EIF1, and RPS12,), protein transport (ATP6V1G1, SLC25A6, and TOMM20), protein folding (HSPA5, HSPA6, and HSPA1B), transcriptional regulation (YY1, ID4, and IER5), and immune response (JUN, AEN, and J o u r n a l P r e -p r o o f 6 RACK1) (FDR < 0.05, Figure 1B , S1F). Importantly, the observed interactions are highly specific for each viral protein, and each protein binds to a precise region within each RNA ( Figures 1C, S1F ).

Using these data, we identified several viral proteins that interact with structural ncRNA components of the spliceosome (U1 and U2 snRNA), the ribosome (18S and 28S rRNA), and the Signal Recognition Particle (7SL) (Figure 1B) . Because these molecular machines are essential for three essential steps of protein production -mRNA splicing, translation, and protein trafficking -we focused on their interactions with viral proteins to understand their functions and mechanisms in SARS-CoV-2 pathogenesis.

After transcription in the nucleus, nascent pre-mRNAs are spliced to generate mature mRNAs which are translated into protein. Splicing is mediated by a complex of ncRNAs and proteins known as the spliceosome. Specifically, the U1 small nuclear RNA (snRNA) hybridizes to the 5' splice site at the exon-intron junction and the U2 snRNA hybridizes to the branchpoint site within the intron to initiate splicing of virtually all human mRNAs (Séraphin et al., 1988) .

We identified a highly specific interaction between the NSP16 viral protein and the U1 and U2 snRNAs ( Figure 1B) . Because U1 and U2 are small RNAs (164 and 188 nucleotides, respectively), we noticed strong enrichment of NSP16-associated reads across the entire length of each. To more precisely define the binding sites, we exploited the well-described tendency of reverse transcriptase to preferentially terminate when it encounters a UV-crosslinked protein on RNA (Konig et al., 2010) (Figures 1A, S1D) . We determined that NSP16 binds to the 5' splice site recognition sequence of U1 (Figures 2A-B, S2A 

Based on the locations of the NSP16 binding sites relative to the mRNA recognition domains of the U1/U2 spliceosomal components, we hypothesized that NSP16 might disrupt splicing of newly transcribed genes ( Figure 2F ). To test this, we co-expressed NSP16 in human cells along with a splicing reporter derived from IRF7 (an exon-intron-exon minigene) fused to GFP (Majumdar et al., 2018) . In this system, if the reporter is spliced, then GFP is made; if not, translation is terminated (via a stop codon present within the first intron) and GFP is not produced ( Figure 3A) . We observed a >3-fold reduction in GFP levels in the presence of NSP16 compared to a control human protein (Figures 3B, S3A ).

To explore whether NSP16 has a global impact on splicing of endogenous mRNAs, we measured the splicing ratio of each gene using nascent RNA sequencing. Specifically, we metabolically labeled nascent RNA by feeding cells for 20 minutes with 5-ethynyl uridine (5EU), purified and sequenced 5EU-labeled RNA, and quantified the proportion of unspliced fragments spanning the 3' splice site of each gene (Figure 3C, S3B) . We observed a global increase in the fraction of unspliced genes in the presence of NSP16 compared to controls ( Figure 3D, S3C,D) .

Given that NSP16 is sufficient to suppress global mRNA splicing, we expect that its expression in SARS-CoV-2-infected cells would result in a global mRNA splicing deficit. To test this, we infected human lung epithelial cells (Calu3) with SARS-CoV-2 and measured splicing levels of newly transcribed mRNAs compared to a mock infected control. As expected, we observed a global increase in the fraction of unspliced transcripts upon SARS-CoV-2 infection, with ~90% of measured genes showing increased intron retention (Figure 3E, S3E ).

Together these results indicate that NSP16 binds to the splice site and branch point sites of U1/U2 to suppress global mRNA splicing in SARS-CoV-2 infected cells ( Figure 3F) . Although NSP16 is known to act as an enzyme that deposits 2'-O-methyl modifications on viral RNAs (Decroly et al., 2011) , our results demonstrate that it also acts as a host virulence factor.

Global disruption of mRNA splicing may act to decrease host protein and mRNA levels by triggering nonsense-mediated decay of improperly spliced mRNAs (Kurosaki et al., 2019) .

Consistent with this, we observed a strong global decrease in steady-state mRNA levels (relative to ncRNA levels) upon SARS-CoV-2 infection ( Figure S3F ). J o u r n a l P r e -p r o o f 8 Inhibition of mRNA splicing suppresses host interferon response to viral infection Because many of the key genes stimulated by interferon (IFN) are spliced, we reasoned that mRNA splicing would be critical for a robust IFN response. To test this, we utilized a reporter line engineered to express alkaline phosphatase upon IFN signaling (mimicking an antiviral response gene). This IFN Stimulated Gene (ISG) reporter line can be stimulated using IFN-β and assayed for reporter induction. We observed strong repression of this IFN responsive gene upon expression of NSP16 ( Figure 3G ) and upon addition of a small molecule that interferes with spliceosomal assembly (Figure S3G ). These results demonstrate that one outcome of NSP16mediated inhibition of mRNA splicing is to reduce the host cells' innate immune response to viral recognition. Consistent with such a role, we observed an increase in intron retention within multiple IFN-responsive genes (such as ISG15 and RIG-I) upon SARS-CoV-2 infection ( Figure   3H , S3H-I).

Once exported to the cytoplasm, spliced mRNA is translated into protein on the ribosome.

Initiation of translation begins with recognition of the 5' cap by the small 40S subunit (which scans the mRNA to find the first start codon). We observed that NSP1 binds exclusively to the 18S ribosomal RNA (Figure 1B and S4A ) -the structural RNA component of the 40S ribosomal subunit.

Several roles for NSP1 have been reported in SARS-CoV and MERS-CoV including roles in viral replication, translational inhibition, transcriptional inhibition, mRNA degradation, and cell cycle arrest (Brockway and Denison, 2005; Kamitani et al., 2009; Lokugamage et al., 2015; Narayanan et al., 2015) . One of the reported roles for NSP1 in SARS-CoV is that it can associate with the 40S ribosome to inhibit host mRNA translation (Kamitani et al., 2009; Tanaka et al., 2012 ), yet it remains unknown whether this association is due to interaction with the ribosomal RNA, protein components of the ribosome, or other auxiliary ribosomal factors. Accordingly, the mechanisms by which NSP1 acts to suppress protein production remain elusive.

J o u r n a l P r e -p r o o f 9 We mapped the location of NSP1 binding to a 37 nucleotide region corresponding to Helix 18 ( Figure 4A) , adjacent to the mRNA entry channel (Simonetti et al., 2020) (Figure 4B ). The interaction would position NSP1 to disrupt 40S mRNA scanning and prevent translation initiation (Figure 4B) , and disrupt tRNA recruitment to the 80S ribosome and block protein production ( Figure S4B) . Interestingly, the NSP1 binding site includes the highly conserved G626 nucleotide which monitors the minor groove of the codon-anticodon helix for tRNA binding fidelity (Ogle et al., 2001) . We noticed that the C-terminal region of NSP1 has similar structural regions to SERBP1 (Brown et al., 2018) and Stm1 (Ben-Shem et al., 2011a) , two known ribosome inhibitors that bind within the mRNA entry channel to preclude mRNA access ( Figure   S4C ). Consistent with this, a recent cryo-EM structure confirms that NSP1 binds to these same nucleotides of 18S within the mRNA entry channel (Thoms et al., 2020) .

Given the location of NSP1 binding on the 40S ribosome, we hypothesized that it could suppress global initiation of mRNA translation. To test this, we performed in vitro translation assays of a GFP reporter in HeLa cell lysates and found that addition of NSP1 led to potent inhibition of translation ( Figure S4D ). We observed a similar NSP1-mediated translational repression when we co-expressed NSP1 and a GFP reporter gene in HEK293T cells (Figure 4C-D) . In contrast, we did not observe this inhibition when we expressed other SARS-CoV-2 proteins (NSP8, NSP9, M) or human proteins (GAPDH) ( Figure 4D ).

To determine if NSP1 leads to translational inhibition of endogenous proteins in human cells, we used a technique called Surface Sensing of Translation (SUnSET) to measure global protein production levels (Schmidt et al., 2009) . In this assay, translational activity is measured by the level of puromycin incorporation into elongating polypeptides ( Figure S4E ). We observed a strong reduction in the level of global puromycin integration in cells expressing NSP1 compared to cells expressing GFP (Figure S4F-G) .

Because NSP1 expression is sufficient to suppress global mRNA translation in human cells, we hypothesized that SARS-CoV-2 infection would also suppress global translation. To test this, we infected a human lung epithelial (Calu3) or monkey kidney (Vero) cell line with SARS-CoV-2 and measured nascent protein synthesis levels using SUnSET. We observed a strong reduction of To explore whether NSP1 binding to 18S rRNA is critical for translational repression, we generated a mutant NSP1 in which two positively charged amino acids (K164 and H165) in the C-terminal domain were replaced with alanine residues (Figure S4C ) (Narayanan et al., 2008) .

We observed a complete loss of in vivo contacts with 18S ( Figure 4G) ; because this mutant disrupts ribosome contact, we refer to it as NSP1∆RC. We co-expressed GFP and NSP1∆RC in HEK293T cells and found that the mutant fails to inhibit translation ( Figure 4H and S4J) . In contrast, mutations to the positively charged amino acids at positions 124/125 do not impact 18S binding ( Figure 4G ) or the ability to inhibit translation ( Figure 4H ).

Together, these results demonstrate that NSP1 binds within the mRNA entry channel of the ribosome and that this interaction is required for translational inhibition of host mRNAs upon SARS-CoV-2 infection.

We explored whether NSP1 binding to 18S rRNA suppresses the ability of cells to respond to IFN-β stimulation upon viral infection. We transfected ISG reporter cells with NSP1, stimulated with IFN-β, and observed robust repression of the IFN responsive gene (>6-fold, Figure 4I ). To confirm that this NSP1-mediated repression occurs in human cells upon activation of double stranded RNA (dsRNA)-sensing pathways typically triggered by viral infection, we treated a human lung epithelial cell line (A549) with poly(I:C), a molecule that is structurally similar to dsRNA and known to induce an antiviral innate immune response (Alexopoulou et al., 2001; Kato et al., 2006) (Figure S4K ). We observed a marked downregulation of IFN-β protein and endogenous IFN-β responsive mRNAs in the presence of NSP1, but not in the presence of NSP1∆RC ( Figure S4L, M) . These results demonstrate that NSP1, through its interaction with 18S rRNA, suppresses the innate immune response to viral recognition ( Figure 4J ).

J o u r n a l P r e -p r o o f 11

Because NSP1 blocking the mRNA entry channel would impact both host and viral mRNA translation, we explored how translation of viral mRNAs is protected from NSP1-mediated translational inhibition. Many viruses contain 5' untranslated regions that regulate viral gene expression and translation (Gaglia et al., 2012) ; all SARS-CoV-2 encoded subgenomic RNAs contain a common 5' leader sequence that is added during negative strand synthesis (Kim et al., 2020b) . We explored whether the leader sequence protects viral mRNAs from translational inhibition by fusing the viral leader sequence to the 5' end of GFP or mCherry reporter genes ( Figure S5A ). We found that NSP1 fails to suppress translation of these leader-containing mRNAs ( Figure 5A -B, S5B). We dissected the leader sequence and found that the first stem loop (SL1) is sufficient to prevent translational suppression upon NSP1 expression ( Figure 5C) or SARS-CoV-2 infection ( Figure 5D ).

We considered three models for how the leader could protect viral mRNAs: (i) it could compete with the ribosome for NSP1 binding, (ii) it could directly recruit free ribosomes or (iii) NSP1 could bind to the leader independently of its ribosome interaction to allosterically modulate the NSP1-ribosome interaction. We reasoned that if the leader competes for NSP1 binding or directly recruits free ribosomes, then the presence of SL1 should be sufficient for protection, regardless of its precise position in the 5' UTR. In contrast, if the leader allosterically modulates ribosome binding then the spacing between the 5' cap (which is bound to NSP1-40S) and SL1 would be critical for protection. To distinguish between these models, we swapped the location of SL1 and SL2 in the 5' leader or inserted 5 nucleotides between the 5' cap and SL1 ( Figure   S5C ) and found that both mutants ablate protection ( Figure 5E, S5D) .

These results indicate that an mRNA requires the 5' leader to be precisely positioned relative to the NSP1-bound 40S ribosome to enable translational initiation ( Figure 5F ). While many aspects of this allosteric model remain to be explored, it would explain how leader-mediated protection can occur on an mRNA only when present in cis. Moreover, this model suggests that NSP1 might also act to further increase viral mRNA translation by actively recruiting the ribosome to its own mRNAs. Consistent with this, we observe a consistent ~20% increase in 12 translation of leader-containing reporter levels upon viral infection ( Figure 5D ) or expression of NSP1 ( Figure S5E ).

Upon engaging the start codon in an mRNA, the 60S subunit of the ribosome is recruited to form the 80S ribosome which translates mRNA. The Signal Recognition Particle (SRP) is a universally conserved complex that binds to the 80S ribosome and acts to co-translationally scan the nascent peptide to identify hydrophobic signal peptides present in integral membrane proteins and proteins secreted from the plasma membrane (Akopian et al., 2013) . When these are identified, SRP triggers ribosome translocation to the endoplasmic reticulum (ER) to ensure proper folding and trafficking of these proteins to the cell membrane (Akopian et al., 2013) .

We identified two viral proteins -NSP8 and NSP9 -that bind at distinct and highly specific regions within the S-domain of the 7SL RNA scaffold of SRP ( Figure 6A , S6A). NSP8 interacts with 7SL in the region bound by SRP54 (the protein responsible for signal peptide recognition, SRP-receptor binding, and ribosome translocation) (Akopian et al., 2013; Holtkamp et al., 2012) ( Figure 6B ). NSP9 binds to 7SL in the region that is bound by the SRP19 protein ( Figure 6B ), which is required for proper folding and assembly of SRP (including proper loading of SRP54) (Akopian et al., 2013) .

Because SRP scans nascent peptides co-translationally, we were intrigued to find that NSP8 also forms a highly specific interaction with 28S rRNA (the structural component of the 60S subunit) ( Figure 6C, S6B) . The binding site on 28S rRNA corresponds to the largest human-specific expansion segment within the ribosome, referred to as ES27 (Parker et al., 2018) . ES27 is highly dynamic, and thus has not been resolved in most ribosome structures (Zhang et al., 2014) . However, when engaged by specific factors, ES27 can become ordered, and was recently shown to be capable of interacting with the ribosome exit tunnel, adjacent to the 60S binding site of SRP ( Figure 6D , S6C) (Wild et al., 2020) .

Together, these observations suggest that NSP8 and NSP9 bind to the co-translational SRP complex. Consistent with this, we find that NSP8 and NSP9 localize broadly throughout the J o u r n a l P r e -p r o o f 13 cytoplasm when expressed in human cells ( Figure S6D ) or upon SARS-CoV-2 infection ( Figure   S6E -F).

Because NSP8 and NSP9 binding on 7SL are positioned to disrupt SRP function, we hypothesized that they may alter translocation of secreted and integral membrane proteins ( Figure S7A ).

To test this, we expressed an SRP-dependent membrane protein (Nerve Growth Factor Receptor, NGFR (Izon et al., 2001a) ) fused via an Internal Ribosome Entry Site (IRES) to a non-membrane GFP ( Figure S7F ). In this system, if a perturbation specifically affects membrane protein levels we expect to see a decrease in the ratio of membrane to non-membrane protein levels. To ensure that the NGFR reporter accurately reports on SRP function, we treated HEK293T cells with siRNAs against SRP54 or SRP19 and found that both lead to a dramatic reduction of the NGFRmembrane protein relative to the non-membrane GFP protein ( Figure S7B) . Similarly, we found that expression of NSP8 and NSP9 (alone or together) lead to a striking reduction in expression of NGFR relative to GFP ( Figure 7A ). Expression of control proteins did not specifically impact NGFR levels ( Figure 7A, S7B) .

To determine if there is a global effect on membrane protein levels, we utilized the SUnSET method to measure puromycin levels in membrane proteins using flow cytometry (see Methods).

We confirmed that disruption of SRP leads to a global reduction in puromycin levels in the cell membrane ( Figure S7C ). We observed a comparable global reduction of puromycin-labeled membrane proteins upon expression of NSP8 or NSP9 individually or together, but not with control proteins (Figure 7B, S7C) .

Because NSP8 and NSP9 are each sufficient to suppress protein integration into the cell membrane, we anticipate that SARS-CoV-2 infection would lead to similar suppression. J o u r n a l P r e -p r o o f 14 However, determining whether SARS-CoV-2 infection specifically impacts membrane protein expression is confounded by the fact that NSP1 inhibits translation of membrane and nonmembrane proteins upon infection.

To address this, we co-expressed a membrane protein reporter (NGFR) containing the 5' viral leader along with a non-membrane GFP reporter containing the viral leader. Upon viral infection, we observed a strong reduction of membrane protein levels ( Figure 7C ), but no reduction in non-membrane GFP levels ( Figure 5D ). To ensure that these effects are specific to SARS-CoV-2 infected cells, we separated individual cells within the infected population into those expressing the viral Spike protein (S+) and those not expressing the protein (S-). We found that the shift in membrane protein levels only occurs in S+ cells ( Figure 7D ), while the Spopulation resembled the mock infected samples ( Figure 7C ). We observed a strong relationship between the level of Spike protein -likely reflecting the amount of viral replication within each cell -and the level of membrane protein suppression ( Figure 7C ). We observed this membrane protein-specific decrease upon infection of human lung epithelial (Calu3, Figure S7D ) and monkey kidney (Vero, Figure 7C -D) cell lines.

Together, these results demonstrate that NSP8 and NSP9 bind to 7SL to disrupt SRP function and suppress membrane protein trafficking in SARS-CoV-2 infected cells. Although NSP8 and NSP9 are thought to be components of the viral replication machinery (Sutton et al., 2004) , our results indicate that they play an additional role as host virulence factors. Because viral membrane proteins also require trafficking to the ER, viral disruption of SRP might negatively impact viral propagation, unless viral proteins are trafficked in an SRP-independent manner ( Figure S7E ) or if NSP8/9 selectively impacts host (but not viral) proteins.

Next we explored how disruption of SRP might be advantageous for viral propagation. Because secretion of IFN and other cytokines is dependent on the SRP complex for secretion ( Figure   S7F ), a central component of the IFN response is dependent on SRP. Accordingly, we hypothesized that NSP8/9-mediated viral suppression of SRP would act to suppress the IFN J o u r n a l P r e -p r o o f 15 response upon infection. To test this, we co-expressed NSP8 and NSP9 and observed a significant reduction in the IFN response relative to a control protein ( Figure S7G ).

Together, these results suggest that SARS-CoV-2 mediated suppression of SRP-dependent protein secretion enables suppression of host immune defenses ( Figure 7E) . Interestingly, many proteins involved in anti-viral immunity -including most cytokines and class I major histocompatibility complex -are membrane-anchored or secreted, and are known to use the SRP pathway for transport (Vermeire et al., 2014) (Figure S7F ), suggesting that there may be other effects of SRP pathway inhibition on SARS-CoV-2 pathogenesis.

We identified several pathogenic functions of SARS-CoV-2 in human cells -including global inhibition of host mRNA splicing, protein translation, and membrane protein trafficking -and described the molecular mechanisms by which the virus acts to disrupt these essential cell processes. Interestingly, all of the viral proteins involved (NSP1, NSP8, NSP9, and NSP16) are produced in the first stage of the viral life cycle, prior to generation of double stranded RNA (dsRNA) products during viral genome replication. Because dsRNA is detected by host immune sensors and triggers the type I interferon response, disruption of these cellular processes would allow the virus to replicate its genome while minimizing the host innate immune response.

Disruption of these three non-overlapping steps of protein production may represent a multipronged mechanism that synergistically acts to suppress the host antiviral response ( Figure 7F) .

Specifically, the IFN response is usually boosted >1,000-fold upon viral detection (through amplification and feedback, Figure S4K ), yet each individual mechanism impacts IFN levels on the order of ~5-10-fold. Accordingly, if each independent mechanism impacts IFN levels moderately, the three together may be able to achieve dramatic suppression of IFN (10 3 =1,000fold). This multi-pronged mechanism may explain the molecular basis for the potent suppression of IFN observed in severe COVID-19 patients.

Interferon is emerging not only as a determinant of disease severity, but also a potential treatment option (Zhou et al., 2020) . As such, our work identifies several therapeutic opportunities for boosting IFN levels upon SARS-CoV-2 infection. For example, disrupting the interaction between NSP1 and 18S rRNA could allow cells to detect and respond to viral infection. Because many small-molecule drugs target ribosomal RNAs (Liaud et al., 2019) , it may be possible to develop drugs to block the NSP1-18S and other interactions. Additionally, disrupting the 5' viral leader may be a potent antiviral strategy since it is critical for translation of all viral proteins. Because SL1 is a structured RNA, it may be possible to design small molecules that specifically bind this structure to suppress viral protein production (Hermann, 2016) . Viral suppression of these cellular functions is not exclusive to the IFN response and will also impact other spliced, translated, secreted, and membrane proteins. Many proteins involved in anti-viral immunity are spliced and/or membrane-anchored or secreted. For example, class I major histocompatibility complex (MHC), which is critical for antigen presentation to CD8 T cells at the cell surface of infected cells (Hansen and Bouvier, 2009 ). By antagonizing membrane trafficking, SARS-CoV-2 may prevent viral antigens from being presented on MHC and allow infected cells to escape T-cell recognition and clearance. In this way, interference with these essential cellular processes might further aid SARS-CoV-2 in evading the host immune response.

More generally, we expect that insights gained from the SARS-CoV-2 protein-RNA binding maps will be critical for exploring additional viral mechanisms. Specifically, we identified many other interactions, including highly specific interactions with mRNAs. For example, NSP12 binds to the JUN mRNA ( Figure S1E ) which encodes the critical immune transcription factor c-Jun which is activated in response to multiple cytokines and immune signaling pathways (Weston and Davis, 2007) . We also identified an interaction between NSP9 and the start codon of the mRNA that encodes COPS5 (Figure 1C) , the enzymatic subunit of the COP9 Signalosome complex which regulates protein homeostasis (Cope and Deshaies, 2003) , suggesting that it might disrupt its translation. Interestingly, COPS5 (also known as JAB1) is known to bind and stabilize c-Jun protein levels (Claret et al., 1996) and several viruses are known to disrupt this protein (Lungu et al., 2008; Oh et al., 2006; Tanaka et al., 2006) . While it remains unknown what, if any, role these interactions play in virally infected cells, the specificity suggests that they may provide a selective advantage for viral propagation.

Together, our results demonstrate that global mapping of RNA binding by viral proteins could enable rapid characterization of mechanisms for emerging pathogenic RNA viruses.

We note several limitations of our current study that will need to be explored in future work. (i)

Our mapping experiments were performed in uninfected human cells expressing tagged viral proteins. Accordingly, it remains possible that our maps may not fully capture all of the interactions that occur when human cells are infected, such as interactions that occur with viralinduced RNAs, in specific viral compartments, or that require multiple viral proteins. (ii) While we characterized the functional and mechanistic roles of several viral proteins and structural ncRNAs, we did not explore what roles viral protein interactions with mRNAs might play. (iii)

How the virus disrupts fundamental cellular processes while still maintaining its own production is still largely undefined. While we showed that the 5' leader is sufficient to relieve translational inhibition by NSP1, we still do not fully understand how this protection occurs and specifically how NSP1 might interact with the viral leader or allosterically modulate ribosome binding.

Similarly, viral membrane proteins are dependent on trafficking to the ER and how NSP8/9 might selectively impact ER translocation of host -but not viral -proteins remains to be explored. (iv) While we showed that viral disruption of these essential cellular functions can suppress IFN, what other roles host cell shutdown might play in viral pathogenesis and in suppressing other aspects of anti-viral immunity, including possible roles in adaptive immune responses, have not been explored. 

The authors declare no competing interests. 

Further information and requests for reagents and resources should be directed to and will be fulfilled by the Lead Contact, Mitchell Guttman (mguttman@caltech.edu).

All constructs and plasmids generated in this study will be made available on request sent to the Lead Contact with a completed Materials Transfer Agreement.

All datasets generated during this study are available at NCBI Short Read Archive: Bioproject PRJNA665692 (viral protein purifications) and PRJNA665581 (nascent and total RNA-Seq) Essential Medium (ATCC) containing 10% FBS and 1% penicillin-streptomycin purchased from Thermo Fisher Scientific. All cell lines were maintained at 37°C under 5% CO 2 . Cells were grown in a humidified incubator at 37ºC with 5% CO 2 .

All experiments using infectious SARS-CoV-2 conducted at the UVM BSL-3 facility were 

Cloning of expression constructs. SARS-CoV-2 protein constructs (with the exception of Nsp11) were a gift from Fritz Roth (see Table S3 for Addgene information) (Kim et al., 2020a) and were LR-cloned (Invitrogen Gateway Cloning, Thermo Fisher Scientific) into mammalian expression destination vector pCAG-Halo-TEV-DEST-V5-IRES-puroR. Note that following LR cloning, proteins were not V5-tagged because all entry clones contained stop codons. For NSP11, an entry clone was generated by BP cloning (Invitrogen Gateway Cloning, Thermo Manganese/Calcium Mix (0.5mM CaCl 2 , 2.5 mM MnCl 2 ). Samples were incubated on ice for 10 minutes to allow lysis to proceed. The lysates were then incubated at 37°C for 10 minutes at 700 rpm shaking on a Thermomixer (Eppendorf). Lysates were cleared by centrifugation at 15,000× g for 2 minutes. The supernatant was collected and kept on ice until bound to the HaloLink Resin (Promega). Of the 1mL lysis volume, 50uL was set aside for input, 20uL used for protein expression confirmation, and the rest for capture on HaloLink Resin as described below. For qPCR analysis, cDNA was generated from purified RNA using Maxima H-reverse transcriptase (Thermo Fisher Scientific) following manufacturer's recommendations.

Amplification reactions were assembled with primer sets indicated in Table S2 and LightCycler® 480 SYBR Green I Master (Roche) following manufacturer's protocols and read out in a Roche Lightcycler 480.

Library construction. RNA-Seq libraries were constructed from purified RNA as previously described (Van Nostrand et al., 2016) . Briefly, after proteinase K elution, the RNA was dephosphorylated (Fast AP) and cyclic phosphates removed (T4 PNK) and then cleaned using Silane beads as previously described (Van Nostrand et al., 2016 ). An RNA adapter containing a RT primer binding site was ligated to the 3' end of the cleaned and end-repaired RNA. The ligated RNA was reverse transcribed (RT) into cDNA, the RNA was degraded using NaOH, and a second adapter was ligated to the single stranded cDNA. Library preparation was the same for input samples except that an initial chemical fragmentation step (90°C for 2 min 30 s in 1X FastAP buffer) was included prior to FastAP treatment. This chemical fragmentation step was designed to be similar to the fragmentation conditions used for purified Halo bound samples. The For staining of infected cells, cells were fixed and permeabilized in 8% formaldehyde 1% Triton, and subsequently labelled with primary antibodies raised in sheep to SARS-CoV-2 at 1/500 dilution, followed by incubation with a rabbit anti-sheep Alexa 555 secondary antibody (Abcam, ab150182) at 1/1000 dilution and mounted with DAPI in the medium (Thermo Fisher Scientific, cat# P36395). Cells were imaged with a Zeiss LSM 880 confocal microscope, with 1 Airy unit pinhole for all primary antibody channel acquisitions and pixel size 0.07 µm x 0.07 µm. The objective lens used was a Zeiss Plan-Apochromatic 63x/1.4NA M27.

Structure modeling NSP1 homology model. The predicted model of SARS-CoV-2 NSP1 was generated using the transform-restrained Rosetta (trRosetta) algorithm, a deep learning-based modeling method based on the Rosetta energy minimization pipeline with additional distance and interaction restraints generated from co-evolution (Yang et al., 2020) . All figures were generated using Pymol (www.pymol.org). NSP1-ribosome model. The model of NSP1 bound to the ribosome was generated using Modeller version 9.24 (Webb and Sali, 2016) . The C-terminal sequence of NSP1 (KHSSGVTRELMRELNGG) was modeled using the structure of SERBP1 bound to the ribosome (PDB ID: 6MTE, chain w) as a template. The default Modeller parameters were used to create an alignment of NSP1 and SERBP1 and to generate the model, and all atoms within 6Å of SERBP1 were included in the model to define the neighboring environment. Twenty models were generated and the model with the lowest DOPE score was selected to visualize with Pymol (Delano, 2002) .

X-ray crystal structures and cryo-electron microscopy structures were obtained from the Protein Data Bank (www.rcsb.org) (Berman et al., 2000) and visualized with PyMOL (Delano, 2002) . For U1 and U2 structural analysis, we used a cryo-EM structure of the pre-catalytic human spliceosome (PDB ID: 6QX9). For 7SL structural analysis, we used an X-ray crystal structure of the human signal recognition particle (PDB ID: 1MFQ). To examine human SRP in the context of the ribosome, we used a cryo-EM structure of the mammalian SRP-ribosome complex (PDB ID: 3JAJ). To analyze the ribosomal ES27 expansion segment, we superimposed a cryo-EM structure of the expansion segment (PDB ID: 6SXO) onto the complete ribosome structure (PDB ID: 3JAJ) using the PyMOL command "super." Finally, for NSP1-18S rRNA structural analysis, we used multiple structures of the ribosome, including structures of the pre-40S subunit (PDB ID: 6G5H), 48S late-stage initiation complex (PDB ID: 6YAL), 80S in complex with SERBP1 (PDB ID: 6MTE), and 80S in complex with Stm1 (PDB ID: 4V88).

J o u r n a l P r e -p r o o f 37 NSP1 was cloned into a bacterial expression vector resulting in N-terminally tagged Halo-6xHistagged Nsp1. The NSP1 sequence was PCR amplified from Addgene Nsp1 entry vector to add a N-terminal 6X HIS tag and restriction enzyme sites for digestion and ligation into N-terminal Halo bacterial expression vector. This construct was transformed into BL21 DE3 E. coli (Agilent), expanded to a 500mL liquid culture, and grown until OD 600 reached 1.0. IPTG was added to a final concentration of 1mM. After 3 hours of IPTG induction, bacteria was centrifuged for 15 min at 5000× g. Pellet was lysed with binding buffer (50mM HEPES, pH 7.5, 20mM MgCl 2 , 600mM NaCl, 2mM TCEP, 10mM Imidazole, 2mM ATP, 1% Triton X-100) supplemented with ATP (2mM), protease inhibitor cocktail (Promega), Benzonase (Sigma) and

Triton-X 100 (Sigma) using 5mL of lysis mix per gram of wet cell paste. Cell suspension was rocked for 20 min at room temperature and then centrifuged at 16,000× g for 20 min at 4°C. Supernatant was incubated with washed iMAC resin (Bio-Rad) and rocked for 20 min at room temperature. We loaded the resin-lysate mixture into an appropriately-sized column and washed with 5 column volumes of binding buffer (50mM HEPES, pH 7.5, 20mM MgCl 2 , 600mM NaCl, 2mM TCEP, 10mM Imidazole, 2mM ATP, 1% Triton X-100) followed by 10 column volumes of wash buffer (50mM HEPES, pH 7.5, 600mM NaCl, 2mM TCEP, 20mM Imidazole, pH 8).

Recombinant NSP1 (rNSP1) was eluted with 5 column volumes of elution buffer by adding 1 column volume at a time with column flow stopped, collecting eluate after each addition, and waiting 15 min between each elution buffer addition. We dialyzed these eluates with a 10mL Spectra-Por® Float-A-Lyzer® G2 (Spectrum Laboratories) into storage buffer (50mM HEPES, pH 7.5, 150mM NaCl, 10% glycerol) at 4°C using 2 exchanges, one after 2 hours and then overnight.

Pierce 1-Step Human Coupled IVT-DNA (Thermo Fisher Scientific) in vitro translation kit was used to measure rNsp1-dependent translation inhibition. Bovine Serum Albumin (BSA), and buffer only controls were used to control for the addition of excess protein or changes in buffer composition. To measure translation inhibition, 5µL in vitro translation reactions were assembled, scaled according to manufacturer's recommendations. The included control plasmid pCFE-GFP was used to measure translational output of the reactions. GFP fluorescence was measured on a BioTek Cytation3 plate reader using emission filters for GFP fluorescence. 1.5µM J o u r n a l P r e -p r o o f 38 stock dilutions of rNsp1 and BSA were made in storage buffer (50mM HEPES, pH 7.5.,150mM NaCl,10% glycerol). Subsequent 10 fold dilutions were made in storage buffer to span a concentration range of 1000 nM to 1 nM for each protein in the final reaction. 10 µL of the diluted protein solution was added to the 5µL translation reactions, and incubated for 5 minutes at room temperature prior to the addition of the GFP reporter plasmid. Duplicate reactions were made to measure variability for each condition. In addition, a buffer only control was included to measure the effect of dilution of the translation reaction by the storage buffer. After the 5 minute incubation, 50 ng of GFP reporter plasmid was added to each reaction and incubated at 30°C for 4 hours prior to fluorescence detection. Two microliters from each reaction was measured in duplicate on a Biotek Cytation3 microplate reader using excitation and emission filters for GFP. Sample readings were blanked by subtracting values obtained from the buffer only control.

Promega's Rabbit Reticulocyte Lysate System was also used to assay translation inhibition. To measure translation inhibition, 10µL in vitro translation reactions were assembled, scaled according to manufacturer's recommendations. For each translation reaction, either 10µL of recombinant protein storage buffer or rNSP1 was added, followed by 500ng of mRNA. After 4 hours of incubation at 30°C, luciferase was read out using the Bright-Glo luciferase assay (Promega) or GFP fluorescence was measured, both on a Biotek Cytation3 plate reader.

We assayed translation in HEK293T cells transfected with mammalian expression vectors, mRNAs, or combinations of these. For mRNA transfections of fluorescence protein translation reporters (including unmodified, +SARS-CoV2 leader sequence, +SL1, +SL2-SL1, and +5nts), DNA templates for in vitro transcription were generated with sequences appended to the 5' end of GFP and mCherry (see Tables S4 and S5 for primers and templates, respectively) and transcribed using HiScribe™ T7 ARCA mRNA Kit with tailing (New England Biolabs). For Nsp1 mRNA transfection, indicated primers from Table S4 were used to add restriction enzyme sites for cloning into pT7CFE1-CHis backbone provided in the Pierce Human 1-step Coupled Acquistion files were analyzed with FlowJo analysis software.

To assay global protein translation, a SUnSET assay was performed as previously described (Schmidt et al., 2009) We transfected these mammalian expression vectors for NSP1 and GFP into HEK293T using BioT transfection reagent. After 3 hours, doxycycline (Sigma) was added to a final concentration of 2µg/mL. After 24 hours, cells were incubated with puromycin (10µg/mL) for 10 min, then washed with fresh media, and harvested with cold PBS. Pelleted cells were lysed for 10 min on J o u r n a l P r e -p r o o f 40 ice (mixing after 5 min) with 100uL RIPA buffer supplemented with protease inhibitor cocktail (Promega). Insoluble debris was pelleted by centrifuging at 12,500 × g for 2.5 minutes and supernatant was run on a Bolt™ 4-12% Bis-Tris Plus Gel (Thermo Fisher Scientific). Proteins were then transferred to nitrocellulose using the iBlot transfer system (Thermo Fisher Scientific) and Western blotting carried out using an anti-puro antibody (clone 12D10, EMD Millipore).

SUnSET in SARS-CoV-2 infection was performed as above with the following modifications.

Cells were infected or not (mock) with SARS-CoV-2, and 48 hpi cells were incubated with puromycin (10µg/mL) for 20 min. Media was aspirated and cells lysed directly in 2X Laemmli's buffer (Biorad), heated at 95ºC for ten minutes and run on a 4-12% NuPAGE Gel (Thermo Fisher Scientific). Proteins were transferred to nitrocellulose using the iBlot transfer system and probed as above.

To assay SRP-dependent membrane protein transport to the cell surface, we monitored surface arrival of exogenously expressed Neuronal Growth Factor Receptor (NGFR) by flow cytometry in the presence of NSPs. Mammalian expression vectors were exchanged for versions that contained an IRES-NGFR to co-express a membrane reporter and thus, for these experiments, LR reactions were carried out with destination vector pB-6xHis-GFP-DEST-IRES-NGFR.

Resulting expression vectors drive protein expression by a dox-inducible promoter, contain the rtTA needed for dox induction, and produce an N-terminally-tagged His-GFP fusion protein and a co-expressed NGFR. The GFP here is an enhanced GFP containing an amino acid substitution (A205K) to generate a monomeric variant based on previous literature (Alberti et al., 2018) .

We transfected these mammalian expression vectors for NSP8, NSP9, NSP1∆RC mutant and 

To knockdown SRP19 and SRP54, siRNAs targeting each (Dharmacon cat# L-019729-01-0005 and L-005122-01-0005, respectively) were transfected into HEK293T cells using Lipofectamine RNAiMAX (Invitrogen) according to manufacturer's protocols. To validate knockdown, transfected cells were assayed by qPCR using primer sets (Table S2 ) to amplify each target as well as normalizer Calm3. Transfections were carried out 48 hours prior to assaying cells, either by qPCR, membrane reporter, or membrane SUnSET (see below) experiments.

Calu3 and Vero cells were transfected with mRNAs encoding leader-NGFR and leader-GFP using TransIT-mRNA Transfection Kit (Mirus) and subsequently infected with SARS-CoV-2 at an MOI of 0.1. After 24 hours, cells were washed with PBS, trypsinized and fixed in 4% PFA for 20 minutes before staining with biotinylated anti-NGFR (BioLegend) and anti-SARS-Cov-2 Spike Antibody (Sino) and subsequently stained with PE-labeled anti-Rabbit (Thermo, P-2771MP) and PacBlue-labeled streptavidin (Thermo, S1222). FACS was performed on a MACSquant Flow cytometer and analyzed using FloJo analysis software; FACS distributions were compared using a 2-tailed Kolmogorov-Smirnov test. For these experiments, RNA was transcribed from a PCR template (see Table S4 ) using the HiScribe T7 ARCA mRNA kit (with tailing).

To assay transport to the cell surface of all plasma membrane proteins, the SUnSET assay was adapted to puro-label surface proteins as previously described (Schmidt et al., 2009) , and read out by flow cytometry. Briefly, cells were incubated with puromycin as described above, followed by two quick washes and a chase with fresh complete media for 50 min. Cells were lifted with 1mM EDTA as described above and stained with an anti-puro antibody (clone 12D10, EMD Millipore) conjugated to Alexa-647. For these experiments, NSP was expressed from the same vector described above for membrane reporter assays. Fluorescence intensity J o u r n a l P r e -p r o o f 42 measurements were taken for GFP and Alexa-647 on a MACSquant Flow cytometer and analyzed using FloJo analysis software; distributions were compared using a 2-tailed Kolmogov-Smirnov.

To assess splicing efficiency, exons 5-6 of mouse IRF7 (ENMUST00000026571.10) containing its endogenous intron were fused upstream of 2A self-cleaving peptide and eGFP and cloned into an MSCV vector (PIG, Addgene) (Mayr and Bartel, 2009 ). This construct was co-transfected into HEK293Ts with NSP16 or GFP and measured 24 hours after transfection by flow cytometry (Macsquant) and analyzed using FloJo analysis software.

SARS-Cov2 or mock infected Calu3 cells and Nsp16-or GAPDH-expressing HEK293Ts were labeled with 5-Ethynyl-uridine (5EU; Jena Bioscience) by adding 5EU containing media to cells for 20 min at a final concentration of 1mM, as previously described (Jao and Salic, 2008) . After the pulse label, cells were washed with warm PBS and lysed in RLT buffer (Qiagen). Total RNA was isolated from cells using manufacturer's protocols for Qiashredder and RNeasy RNA isolation (both Qiagen), followed by Turbo DNase treatment (Ambion, Thermo Scientific), and Zymo RNA Clean and Concentrate. For each sample, 2µg of RNA was used for ligation of a unique barcoded RNA adaptor, following the relevant steps in the protocol described above in Library Construction of RNA-seq libraries. Samples were then pooled before proceeding to biotinylation steps.

To biotinylate 5EU-labeled RNA, samples were first mixed, in order, with water, HEPES (100 mM), biotin picolyl azide (1 mM; Click Chemistry Tools) and Ribolock RNase inhibitor, then added to premixed CuSO 4 (2 mM) and THPTA (10mM), and finally added to freshly prepared J o u r n a l P r e -p r o o f 43 sodium ascorbate (12mM), as previously described (Hong et al., 2009) . The click reaction was incubated for 1 hour at 25ºC with 1000rpm shaking on an Eppendorf thermomixer followed by RNA purification using >17nt protocol for Zymo Clean and Concentrate.

We completed three rounds of sequential capture on streptavidin beads to isolate nascent transcripts (see Figure S3B ). To capture biotinylated RNA, MyOne Streptavidin C1 Dynabeads (ThermoFisher Scientific) were first washed three times in Urea buffer (10mM HEPES, pH 7.5, 10mM EDTA, 0.5M LiCl, 0.5% Triton X-100, 0.2% SDS, 0.1% sodium deoxycholate, 2.5mM TCEP, 4M Urea) followed by three additional washes in M2 buffer (20mM Tris, pH 7.5, 50mM

NaCl, 0.2% Triton X-100, 0.2% sodium deoxycholate, 0.2% NP-40). Washed beads were mixed with 3 parts 4M Urea buffer and 1 part biotinylated RNA and incubated for 60 min with 900rpm thermomixer shaking at room temperature.

After magnetic separation, beads were washed 3 times with M2 buffer followed by 3 washes with Urea buffer at 37 ºC at 750rpm for 5 min. RNA was eluted from beads in 2 sequential elutions by incubating with elution buffer (5.7M guanidine thiocyanate , 1% N-lauroylsarcosine;

both Sigma) at 65 ºC for 2 minutes, repeating with more elution buffer for a second elution.

The elutions were pooled, diluted with Urea buffer, incubated with pre-washed streptavidin beads, washed, and eluted for 2 additional rounds exactly as described above for a total of 3 sequential captures.

Final elutions were pooled, cleaned with Zymo RNA Clean and Concentrate following manufacturer's protocols, and carried through RNA-seq library preparation as described above starting with the reverse transcription step.

HEK-Blue™ ISG cells were seeded in 96 well plates, transfected with Nsp1 mammalian expression vectors using BioT and stimulated with 50 ng/ml human IFN-B (R&D Systems).

Supernatants were assayed for alkaline phosphatase as per manufacturer instructions using J o u r n a l P r e -p r o o f 44 QUANTI-Blue reagent (Invivogen).

HEK-293T cells were seeded in 6 well plates, transfected with either Halo-tagged GapdH, Nsp1, NSP8 and NSP9 in combination, or NSP16 mammalian expression vectors using BioT. 24 hours later, the media was replaced with media containing 50 ng/ml human IFN-β (R&D Systems).

Expression was assayed using live cell Halo-imaging. Halo-TMR ligand was diluted 1:200 in media and added to the culture for a 1:1000 final dilution. Samples were incubated 30 minutes at 37°C, 5% CO 2 and then the media was aspirated. Wells were rinsed twice with PBS, then media was added back to the wells. Samples were incubated 30 minutes at 37°C, 5% CO 2 to allow uncoupled ligand to diffuse out of the cells. Media was then aspirated and replaced, and cells were imaged by widefield fluorescence microscopy. Cultures were ultimately harvested for RNA 24 hours later, or 48 hours post transfection.

A549s were seeded in 6 well plates, transfected with NSP1 mammalian expression vectors using Lipofectamine 2000 and stimulated with 1 µg/ml HMW poly(I:C) (Invivogen) 24h after transfection. Supernatant was assayed for secreted IFN-β by ELISA (Human IFN Beta ELISA, High Sensitivity, PBL) 24 hours after stimulation, and RNA from cells was purified and assessed for ISG gene expression as normalized to GAPDH expression (SYBR Green Master Mix, Bio-Rad). Primers used for qPCR are listed in Table S2 .

Sars-CoV-2 Leader sequence was appended to the 5' end of GFP and mCherry reporter templates via PCR. PCR templates were then transcribed using HiScribe T7 ARCA mRNA kit (with tailing). Leader mutants, including SL1 only, SL1/SL2 swap, and +5nts mutants were likewise appended to the 5' end of fluorescent reporter templates via PCR and transcribed using Hiscribe T7 ARCA kit. mRNA reporters were transfected in HEK-293T cells with Lipofectamine MessengerMax. To measure fluorescence of mCherry and GFP reporters, 24 hours post transfection cells were either lifted with PBS and transferred into black 96 well plates for fluorescence readout on a Biotek Cytation 3 or trypsinized and processed for flow cytometry. 

Sequence alignment and analysis. For Halo purifications and RNA binding mapping sequencing reads were aligned to a combined genome reference containing the sequences of structural RNAs (ribosomal RNAs, snRNAs, snoRNAs, 45S pre-rRNA) and annotated mRNAs (RefSeq hg38) using Bowtie2. To distinguish between the nascent pre-ribosomal RNA and mature 18S, 28S, and 5.8S rRNA, we separated each of the components of the 45S into separate sequence units for alignment (e.g. ITS, ETS). We excluded all low quality alignments (MAPQ < 2) from the analysis. For mRNA analysis, we removed PCR duplicates using the Picard MarkDuplicates function (https://broadinstitute.github.io/picard/).

For each RNA, we enumerated 100 nucleotide windows across the entire RNA. For each window, we calculated the enrichment by computing the number of reads overlapping the window in the protein elution sample divided by the total number of reads within the protein elution sample. We normalized this ratio by the number of reads in the input sample divided by the total number of reads in the input sample. Because all windows overlapping a gene should have the same expression level in the input sample (which represents RNA expression), we estimated the number of reads in the input as the maximum of either (i) the number of reads over the window or (ii) the median read count over all windows within the gene. This approach provides a conservative estimation of enrichment because it prevents windows from being scored as enriched if the input values over a given window are artificially low, while at the same time accounting for any non-random issues that lead to increases in read counts over a given window (e.g. fragmentation biases or alignment artifacts leading to non-random assignment or pileups). J o u r n a l P r e -p r o o f 46 We calculated a multiple testing corrected p-value using a scan statistic, as previously described (Guttman et al., 2009 (Guttman et al., , 2010 . Briefly, n was defined as the number of reads in the protein elution plus the number of reads in the control sample. p was defined as the total number of reads in the protein elution sample divided by the sum of the protein elution sample total reads and total reads in the control sample. w was the size of the window used for the analysis (100 nucleotides). The scan statistic p-value was defined using the Poisson estimations based on standard distributions previously described (Naus, 1982) .

Because RNA within input samples are fragmented differently than the protein elution samples, we noticed that the overall positional distribution of protein elution samples was distinct from Input distributions. Accordingly, we used the remaining protein elution samples (rather than Input) as controls for each protein. Specifically, this enabled us to test whether a given protein is enriched within a given window relative to all other viral and control proteins. Enrichments were computed as described above. These values are plotted in Figure 1 and Table 1 .

IGV (Robinson et al., 2011) and were generated by either: (i) computing the enrichment for each nucleotide as described above. In this case, the read count for each nucleotide was computed as the total number of reads that overlapped the nucleotide. (ii) Counting the number of RT stop sites at a given nucleotide. In this case, we compute the alignment start position of the second in pair read and computed a count of each nucleotide. We normalized this count by the total number of reads in the sample to account for sequencing depth generated. We then normalized this ratio by the same ratio computed for the control sample (merge of all other protein samples) for each nucleotide.

Heatmaps were generated using Morpheus (https://software.broadinstitute.org/morpheus/). All values were included if they contained a significant 100nt window with a p-value<0.001 (see above) and minimum enrichment of 3-fold above the control sample.

Gene ontology analysis. The 66 non-N enriched mRNAs were analyzed against the Gene Ontology Biological Processes and Reactome gene sets using the Molecular Signatures Database (MSigDB) (Liberzon et al., 2015) . Significantly enriched gene sets with an FDR<0.05 were used.

J o u r n a l P r e -p r o o f 47 To ensure that significant gene sets were not being driven by the multiple ribosomal proteins or histone proteins, these analyses were also carried out excluding these proteins.

Sequenced reads were demultiplexed according to barcoded RNA adaptor sequences ligated to each respective sample. Trimmomatic (https://github.com/timflutre/trimmomatic ) was used to remove any contaminating Illumina primer sequences in the reads and low quality reads.

Demultiplexed and trimmed files were then aligned to a hg19 reference genome using the spliceaware STAR aligner (https://github.com/alexdobin/STAR). Alignments were then deduplicated for PCR duplicates using PICARD MarkDuplicates (https://broadinstitute.github.io/picard/).

Aligned read-fragments were defined as read1 and read2 contained within a paired-end read fragment along with the insert between these two reads. We defined a set of high-quality represented isoforms per gene using the APPRIS database (Rodriguez et al., 2013) . All readfragments that spanned any 3' splice site within an isoform of one of these genes was retained.

For each 3' splice site spanning fragment, we classified the read-fragment as a spliced fragment if it spanned an exon-exon junction (e.g. aligned entirely within 2 distinct exons) or an unspliced fragment if it spanned an intron-exon junction (e.g. one of the reads was contained -or partially contained -within the intron). For each isoform, we computed an unspliced ratio by counting the total number of reads that were classified as unspliced divided by the total number of readfragments spanning 3' splice sites within that gene. To ensure that the splicing ratio that we measured is a reliable metric and not inflated/deflated due to low read counts, we only included genes that contained at least 10 read-fragments in each sample and where the total number of reads in the control and sample conditions (when merged together) contained a significant number of reads to reliably measure a difference between the two groups as measured by a hypergeometric test (p<0.01).

Because different genes contain different baseline splicing ratios due to gene length and coverage, we computed a change in the splicing ratio for each gene independently. To do this, we subtracted the unspliced ratio for each sample from the average unspliced ratio for that gene J o u r n a l P r e -p r o o f 48 in all of the control samples. We plotted the overall distribution of these differences in splicing ratios as violin plots for each sample. If there is no change in splicing ratio, we would expect that some genes would have higher splicing ratios and others lower splicing ratios but that the overall distribution would be centered around 0.

Total RNA-Seq libraries were generated from the same mock infected and SARS-CoV-2 virally infected Calu3 samples treated with 5EU. Prior to 5EU purification, total RNA was taken and an RNA-Seq library constructed as described above using barcoded RNA adapters. Cytoplasmic ribosomal RNAs (18S and 28S) were depleted using NEBNext ribosomal RNA depletion kit (NEB E6310L) per manufacturers recommendations. Demultiplexed reads were aligned using Bowtie2 (http://bowtie-bio.sourceforge.net/bowtie2/index.shtml) to custom genomes encoding classical noncoding RNAs (ncRNAs) or human messenger RNAs (mRNAs). Expression levels were computed for each mRNA by counting the total number of sequencing reads aligned to the mature mRNA. To normalize across the different libraries, we computed the read counts for each sample that align to non-spliced structural non-coding RNAs -excluding rRNA but including snRNAs, 7SL, 7SK, etc. We then divided each mRNA count by the sum of all ncRNA counts. This normalized value for each gene per sample was then converted into a fold-change by dividing this normalized value to the mean value for both mock infected samples. The fold change of each gene relative to mock was plotted across all mRNAs as a violin plot. and enrichment >3-fold for any of the viral proteins is reported.

• NSP16 binds mRNA recognition domains of U1/U2 snRNAs and disrupts mRNA splicing • NSP1 binds in the mRNA entry channel of the ribosome to disrupt protein translation • NSP8 and NSP9 bind the Signal Recognition Particle and disrupt protein trafficking • These disruptions to protein production suppress the interferon response to infection

In brief -SARS-CoV-2 proteins directly engage host RNAs and dysregulate RNA-based processes to suppress the interferon response J o u r n a l P r e -p r o o f 

); Office of the Vice President for Research at UVM

M.G. is a NYSCF-Robertson Investigator

Signal Recognition Particle: An Essential Protein-Targeting Machine

A User's Guide for Phase Separation Assays with Purified Proteins

Recognition of doublestranded RNA and activation of NF-κB by Toll-like receptor 3

Visualizing late states of human 40S ribosomal subunit maturation

The proximal origin of SARS-CoV-2

Sars-cov-2 (Covid-19) by the numbers

Innate immune evasion strategies of DNA and RNA viruses

The structure of the eukaryotic ribosome at 3.0 Å resolution

The structure of the eukaryotic ribosome at 3.0 Å resolution

The Protein Data Bank

Mutagenesis of the murine hepatitis virus nsp1-coding region identifies residues important for protein processing, viral RNA synthesis, and viral replication

Structures of translationally inactive mammalian ribosomes. Elife 7

MERS-CoV 4b protein interferes with the NF-κB-dependent innate immune response during infection

Mechanism of 5' splice site transfer for human spliceosome activation

Genomics functional analysis and drug screening of SARS-CoV-2

RIG-I and Other RNA Sensors in Antiviral Immunity

A new group of conserved coactivators that increase the specificity of AP-1 transcription factors

COP9 signalosome: A multifunctional regulator of SCF and other cullin-based ubiquitin ligases

Crystal structure and functional analysis of the SARS-coronavirus RNA cap 2′-o-methyltransferase nsp10/nsp16 complex

PyMOL molecular graphics system

The coding capacity of SARS-CoV-2

A Common Strategy for Host RNA Degradation by Divergent Viruses

MINIREVIEW Recombination, Reservoirs, and the Modular Spike: Mechanisms of Coronavirus Cross-Species Transmission

Chromatin signature reveals over a thousand highly conserved large non-coding RNAs in mammals

Ab initio reconstruction of cell type-specific transcriptomes in mouse reveals the conserved multi-exonic structure of lincRNAs

Impaired type I interferon activity and inflammatory responses in severe COVID-19 patients

MHC class i antigen presentation: Learning from viral evasion strategies

Small molecules targeting viral RNA

Dynamic switch of the signal recognition particle from scanning to targeting

Analysis and optimization of coppercatalyzed azide-alkyne cycloaddition for bioconjugation

Notch1 regulates maturation of CD4+ and CD8+ thymocytes by modulating TCR signal strength

Notch1 regulates maturation of CD4+ and CD8+ thymocytes by modulating TCR signal strength

Exploring RNA transcription and turnover in vivo by using click chemistry

Sensing of RNA Viruses: a Review of Innate Immune Receptors Involved in Recognizing RNA Virus Invasion

A twopronged strategy to suppress host protein synthesis by SARS coronavirus Nsp1 protein

Differential roles of MDA5 and RIG-I helicases in the recognition of RNA viruses

Poliovirus 2APro Increases Viral mRNA and Polysome Stability Coordinately in Time with Cleavage of eIF4G

A flexible genome-scale resource of SARS-CoV-2 coding sequence clones

The Architecture of SARS-CoV-2 Transcriptome

Inducible transgene expression in human iPS cells using versatile all-in-one piggybac transposons

Multiple Anti-Interferon Actions of the Influenza A Virus NS1 Protein

iCLIP reveals the function of hnRNP particles in splicing at individual nucleotide resolution

Induced structural changes of 7SL RNA during the assembly of human signal recognition particle

Quality and quantity control of gene expression by nonsense-mediated mRNA decay

Cellular response to small molecules that selectively stall protein synthesis by the ribosome

The Molecular Signatures Database Hallmark Gene Set Collection

Translational control by viral proteinases

Middle East Respiratory Syndrome Coronavirus nsp1 Inhibits Host Gene Expression by Selectively Targeting mRNAs Transcribed in the Nucleus while Sparing mRNAs of Cytoplasmic Origin

Type I interferon susceptibility distinguishes SARS-CoV-2 from SARS-CoV

HaloTag: A novel protein labeling technology for cell imaging and protein analysis

Down-regulation of Jab1, HIF-1Α, and VEGF by Moloney murine leukemia virus-ts1 infection: A possible cause of neurodegeneration

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

Programmed Delayed Splicing: A Mechanism for Timed Inflammatory Gene Expression

Widespread Shortening of 3′UTRs by Alternative Cleavage and Polyadenylation Activates Oncogenes in Cancer Cells

Human cytomegalovirus inhibits major histocompatibility complex class II expression by disruption of the Jak/Stat pathway

Inhibition of Stress Granule Formation by Middle East Respiratory Syndrome Coronavirus 4a Accessory Protein Facilitates Viral Translation, Leading to Efficient Virus Replication

Severe Acute Respiratory Syndrome Coronavirus nsp1 Suppresses Host Gene Expression, Including That of Type I Interferon, in Infected Cells

Coronavirus nonstructural protein 1: Common and distinct functions in the regulation of host and viral gene expression

Approximations for Distributions of Scan Statistics

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

Recognition of cognate transfer RNA by the 30S ribosomal subunit

Jab1 mediates cytoplasmic localization and degradation of West Nile virus capsid protein

The Expansion Segments of 28S Ribosomal RNA Extensively Match Human Messenger RNAs

Integrative genomics viewer

APPRIS: Annotation of principal and alternative splice isoforms

SUnSET, a nonradioactive method to monitor protein synthesis

A U1 snRNA:pre-mRNA base pairing interaction is required early in yeast spliceosome assembly but does not uniquely define the 5' cleavage site

Role of Nonstructural Proteins in the Pathogenesis of SARS-CoV-2

Structural Insights into the Mammalian Late-Stage Initiation Complexes

RNA-RNA and RNA-protein interactions in coronavirus replication and transcription

Type I Interferons in Host Defense

The nsp9 Replicase Protein of SARS-Coronavirus

Severe Acute Respiratory Syndrome Coronavirus nsp1 Facilitates Efficient Propagation in Cells through a Specific Translational Shutoff of Host mRNA

The hepatitis B virus X protein enhances AP-1 activation through interaction with Jab1

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

Signal Peptide-Binding Drug as a Selective Inhibitor of Co-Translational Protein Translocation

Structures of the scanning and engaged states of the mammalian srp-ribosome complex

The genetic sequence, origin, and diagnosis of SARS-CoV-2

Comparative protein structure modeling using MODELLER

The JNK signal transduction pathway

MetAP-like Ebp1 occupies the human ribosomal tunnel exit and recruits flexible rRNA expansion segments

Recognition of viruses by cytoplasmic sensors

Improved protein structure prediction using predicted interresidue orientations

Influenza Virus NS1 Protein-RNA Interactome Reveals Intron Targeting

Viral and host factors related to the clinical outcome of COVID-19

Structural basis for interaction of a cotranslational chaperone with the eukaryotic ribosome

Interferon-α2b Treatment for COVID-19

SARS-CoV-2 viral load in upper respiratory specimens of infected patients

We thank Fritz Roth for clones; Marko Jovanovic, Bil Clemons, Shu-ou Shan, and Jamie