The SARS-CoV-2 protein NSP2 impairs the silencing capacity of the human 4EHP-GIGYF2 complex

Summary There is an urgent need for a molecular understanding of how SARS-CoV-2 influences the machineries of the host cell. Herein, we focused our attention on the capacity of the SARS-CoV-2 protein NSP2 to bind the human 4EHP-GIGYF2 complex, a key factor involved in microRNA-mediated silencing of gene expression. Using in vitro interaction assays, our data demonstrate that NSP2 physically associates with both 4EHP and a central segment in GIGYF2 in the cytoplasm. We also provide functional evidence showing that NSP2 impairs the function of GIGYF2 in mediating translation repression using reporter-based assays. Collectively, these data reveal the potential impact of NSP2 on the post-transcriptional silencing of gene expression in human cells, pointing out 4EHP-GIGYF2 targeting as a possible strategy of SARS-CoV-2 to take over the silencing machinery and to suppress host defenses.

Owing to the urgent need to better understand SARS-CoV-2 biology, several CRISPR-Cas9 and proteomicbased screening campaigns have investigated the landscape of host factors which are targeted by the virus proteome (Hoffmann et al., 2021;Sadegh et al., 2020;Gordon et al., 2020aGordon et al., , 2020bDavies et al., 2020). These screens identified more than 300 high-confidence protein-protein interactions between human and SARS-CoV-2 proteins, highlighting the intimate connection of SARS-CoV-2 proteins with multiple biological processes, including protein trafficking, transcription, and mRNA translation. Among these interactions, the non-structural protein 2 (NSP2) has been found to interact with key host proteins involved in vesicle trafficking (FKBP15 and WASHC) and mRNA translation (4EHP and GIGYF2) which could be of therapeutic importance (Gordon et al., 2020b;Davies et al., 2020).
NSP2 exists in all coronaviruses studied to date, including SARS-CoV-1, SARS-CoV-2, MERS, and in their closely related b-CoVs infecting mammals. Although its role in the SARS-CoV-2 pathogenicity has not been fully elucidated, the deletion of NSP2 in SARS-CoV-1 attenuates viral growth and RNA synthesis (Graham et al., 2005). Recent findings showed that SARS-CoV-2 NSP2 is undergoing positive nature selection and could be thus essential to the virus (Angeletti et al., 2020;Flores-Alanis et al., 2021). At the structural level, NSP2 has a complex multi-domain topology including an N-terminal domain with a highly conserved zinc binding site, and a C-terminal region rich in b-strands. With the exception of the zinc binding site, NSP2 displays a rapidly evolving surface with the presence of natural variations that could impact host-virus interactions (Ma et al., 2021;Mompean et al., 2021;Gupta et al., 2021;Slavin et al., 2021). Among its host interactors, SARS-CoV-2 NSP2 interacts with the 4EHP-GIGYF2 complex, a key machinery in translational silencing and mRNA decay (Davies et al., 2020;Cornillez-Ty et al., 2009;Gordon et al., 2020aGordon et al., , 2020b. It is worth noting that the NSP2/4EHP-GIGYF2 association is conserved across SARS-CoV-1 and MERS, corroborating a functional importance for b-CoV infection in general (Gordon et al., 2020a;Cornillez-Ty et al., 2009).
The cap-binding eIF4E-Homologous Protein (4EHP) is an integral component of post-transcriptional silencing mechanisms through competing with the eIF4F complex for binding to the mRNA 5 0 cap (Christie and Igreja, 2021). In complex with the GRB10-interacting GYF (glycine-tyrosine-phenylalanine domain) protein 2 (GIGYF2), the cap-binding activity of 4EHP is required for the optimal translational repression by mi-croRNAs (miRNAs), as well as the RNA-binding proteins ZNF598 and tristetraprolin (TTP) (Morita et al., 2012;Fu et al., 2016;Villaescusa et al., 2009;Cho et al., 2005Cho et al., , 2006Chapat et al., 2017b;Chen and Gao, 2017). In the case of miRNA-driven silencing, the recruitment of 4EHP-GIGYF2 is initiated by the miRNAinduced silencing complex (miRISC), an assembly of argonaute and TNRC6/GW182 proteins. 4EHP can be physically mobilized through the interaction of the GYF domain of GIGYF2 with a proline-prolineglycine-leucine (PPGL) motif in TNRC6/GW182 (Kryszke et al., 2016;Schopp et al., 2017). At the functional level, this 4EHP/miRNA axis is required to control ERK signaling, as well as to suppress IFN-b production by affecting the miR-34a-induced translational silencing of Ifnb1 mRNA Jafarnejad et al., 2018).
Beyond miRNA action, 4EHP-GIGYF2 also forms a translation inhibitory complex with the RNA-binding protein ZNF598, which functions in ribosome stalling on internally polyadenylated mRNAs during ribosome quality control (Morita et al., 2012;Garzia et al., 2017). ZNF598 is also necessary for the repression of TTPtargeted mRNAs that encode inflammatory cytokines. The 4EHP-GIGYF2-ZNF598 complex binds TTP during an innate immune response in mouse macrophages to control the production of TTP-targeted mRNAs such as TNF-a, Ier3, Csf2, and Cxcl10. In all cases, TNRC6/GW182, ZNF598, and TTP display a comparable binding mode to GIGYF2, namely via the recognition of a proline stretch by the GYF domain of GIGYF2 (Tollenaere et al., 2019;Fu et al., 2016).
A recent genetic screen has revealed that both 4EHP and GIGYF2 are necessary for infection by SARS-CoV-2 in vitro, while dispensable for seasonal coronaviruses (Hoffmann et al., 2021). The contribution of 4EHP-GIGYF2 into the pathogenicity of SARS-CoV-2 could therefore originate from their interaction with NSP2. In the present report, we focused on the physical and functional interplays existing between the 4EHP-GI-GYF2 complex and the SARS-CoV-2 protein NSP2 in human cells. Combining interaction assays and reporter-based approaches, our data shed light on the negative impact of NSP2 on the 4EHP-GIGYF2-mediated translational silencing of gene expression.

NSP2 binds the 4EHP-GIGYF2 complex in cellulo
Early large-scale studies reported the capacity of SARS-CoV-2 NSP2 to bind the 4EHP-GIGYF2 complex using affinity purification-mass spectrometry (Gordon et al., 2020a(Gordon et al., , 2020b. To validate the physical association of NSP2 with the GIGYF2-4EHP complex, we first sought to detect their interaction using co-immunoprecipitation (co-IP). An inducible Flag-tagged version of NSP2 was stably expressed in HEK293 Flp-In T-REX cells and co-IPs were performed following tetracycline-induced expression of Flag-NSP2. Lysates prepared from control, non-induced, and induced cells were immunoprecipitated with anti-Flag antibody. Subsequent Western blot (WB) analysis of the co-IP fraction showed that Flag-NSP2 efficiently binds both endogenous GIGYF2 and 4EHP ( Figure 1A). This interaction was similarly detected in RNase A-treated lysates, indicating that the NSP2/4EHP-GIGYF2 interaction occurs in an RNA-independent manner. The GIGYF2-associated protein ZNF598 was also found along with NSP2, while CNOT9, a subunit of the CCR4-NOT complex known to bind GIGYF2, was not detected (Ajiro et al., 2009) (Figure 1A). Together, these data demonstrate that NSP2 physically associates with the 4EHP-GIGYF2 complex, as well as their interacting protein ZNF598.
To elucidate which protein is involved in the formation of this complex, GIGYF2-and 4EHP-knockout (KO) HEK293 cells were used to conduct co-IPs with Flag-NSP2. Vectors expressing Flag-NSP2, or Flag as a control, were then transiently transfected in these KO populations, as well as in their wild-type (WT) counterpart. Following Flag IP in the 4EHP KO cells, we observed that the NSP2/GIGYF2 interaction was still detectable, while ZNF598 co-IP was reduced, indicating a plausible contribution of 4EHP in NSP2 binding. By iScience Article contrast, in the absence of GIGYF2, we could not detect any interaction of NSP2 with either 4EHP or ZNF598 ( Figure 1B), supporting a central role of GIGYF2 in NSP2 binding. However, a contribution of 4EHP cannot be excluded since its level is decreased in the GIGYF2 KO cells due to a co-stabilization effect ( Figure 1B) (Morita et al., 2012).
We then performed immunofluorescence staining and confocal imaging to analyze the subcellular distribution of NSP2. Following transient expression of Flag-tagged NSP2 in WT and GIGYF2 KO cells, immunofluorescence staining was performed with antibodies raised against the Flag sequence and endogenous GIGYF2. In WT cells, a diffuse cytoplasmic signal was detected for both Flag-tagged NSP2 and GIGYF2 ( Figure S1A), indicating their mutual localization in the cytoplasm. By contrast, the GIGYF2 signal was lost in GIGYF2 KO cells while the distribution of Flag-NSP2 remained unchanged ( Figure S1A), confirming the specificity of the anti-GIGYF2 antibody. The subcellular localization of NSP2/GIGYF2 interaction was then examined using an in situ proximity ligation assay (PLA). PLA was conducted in Flag-NSP2-expressing cells using the anti-Flag and anti-GIGYF2 antibodies. Following confocal imaging, a spot-like signal was abundantly detected in the cytoplasm of HEK293T cells expressing Flag-NSP2, indicating the spatial proximity between endogenous GIGYF2 and Flag-NSP2 ( Figure 1C). This PLA signal was not observed in cells transfected with a control plasmid. Similarly, the PLA signal showed a significant reduction when the Flag/ GIGYF2 staining was performed in GIGYF2 KO cells expressing Flag-NSP2 (Figures S1B and S1C), supporting the specificity of the interaction. iScience Article In addition to its diffuse cytoplasmic localization, GIGYF2 is known to be found in P-bodies, a subclass of RNA granules enriched in translationally repressed mRNAs and silencing factors (Amaya Ramirez et al., 2018). We therefore tested whether NSP2 could also be detected in P-bodies by examining the localization of Flag-tagged NSP2 alongside DDX6, a known resident of P-bodies. Using immunofluorescence microscopy, we found that Flag-NSP2 did not form foci that co-localized with DDX6 in HEK293T cells (Figure S1D). Intensity line scans were performed along the DDX6 foci to confirm the absence of relationship between the spatial distribution of NSP2 and P-bodies, thus indicating that NSP2 could preferentially bind the diffuse form of GIGYF2 instead of its P-body-associated counterpart. To confirm this point, we took advantage of the properties of biotinylated isoxazole (b-isox), a compound which selectively precipitates proteins located in RNA granules such as P-bodies and stress granules (Han et al., 2012;Kato et al., 2012) ( Figure S1E). Extracts of HEK293T expressing Flag-NSP2 were exposed to 100 mM of b-isox, or DMSO as a mock control. By comparing the level of proteins left in the soluble fraction after precipitation (unbound) with the precipitated fractions (pellet), we confirmed that Flag-NSP2 was not recovered in the pellet whereas a significant proportion of endogenous GIGYF2 was selectively enriched in the b-isox precipitate, alongside DDX6 ( Figure S1F). Since NSP2 could not be precipitated by b-isox along with GIGYF2, we concluded that it only binds the diffuse form of GIGYF2 in the cytoplasm.
We next investigated which part of NSP2 binds the 4EHP-GIGYF2 complex. For this purpose, we generated a collection of vectors encoding Flag-tagged truncated versions of NSP2 which were expressed in HEK293T for Flag IP ( Figures 1D and 1E). Incremental terminal deletions revealed that the N-terminal half of NSP2 is required for the maximal co-IP of the endogenous 4EHP-GIGYF2 complex as well as ZNF598. While weakly expressed, the NSP2 1À350 fragment remained the minimal segment which was able to bind 4EHP-GIGYF2. Interestingly, deleting either the N-terminal extremity (fragment NSP2 107À350 ) or the middle region (fragment NSP2 1À212 ) impaired the NPS2/4EHP-GIGYF2 interaction, indicating a large interaction surface between 4EHP-GIGYF2 and an intact N-terminal half of NSP2 ( Figure 1E).

The NSP2/GIGYF2-4EHP interaction involves multiple binding sites
We then sought to delineate which part of the 4EHP-GIGYF2 complex is targeted by NSP2. Human GIGYF2 is a 150 kDa scaffolding protein composed of several domains interspaced by intrinsically disordered regions. These include the 4EHP-binding domain at the N-terminus, the so-called GYF domain, a putative single alpha-helix (SAH), and many glutamine-rich stretches (polyQ) at the C-terminus ( Figure 2A) (Peter et al., 2017;Suveges et al., 2009;Morita et al., 2012;Kofler and Freund, 2006). V5-tagged fragments of GIGYF2 were designed to isolate these features, and expressed along with Flag-NSP2 for Flag IP (Figures 2A and 2B). Our WB analysis of the IP fractions revealed that NSP2 binds two distinct segments of GIGYF2, namely the central putative SAH region (residues: 743-1,085), and to a smaller extent, the N-terminal extremity containing the 4EHP-binding motif (residues: 1-267; Figure 2B). The strong interaction between NSP2 and GIGYF2 743À1,085 prompted us to test their direct association. For this, full-length hexahistidine (His 6 )-tagged NSP2 and glutathione S-transferase (GST)-fused GIGYF2 743À1,085 were recombinantly expressed in Escherichia coli (E. coli) and individually purified to perform a His pull-down assay. Because the untagged GIGYF2 743À1,085 fragment appeared to be unstable following our two-step purification process, we tested whether this recombinant protein could be recognized by an anti-GIGYF2 antibody raised against the 756-1,104 region. Our Western blot assay confirmed that a predominant $42-kDa band corresponding to an intact GIGYF2 743À1,085 is specifically detected by the anti-GIGYF2 antibody (Figure S2A). Following incubation of His 6 -NSP2 with untagged GIGYF2 743À1,085 on a Ni-NTA resin, analysis of the eluates by SDS-PAGE and Coomassie blue staining revealed a specific retention of GIGYF2 743À1,085 with His 6 -NSP2, confirming their direct interaction in vitro ( Figure 2C). Western blot analysis of these samples confirmed that the retained band is specifically recognized by the anti-GIGYF2 antibody ( Figure S2B). GST alone was used as control and did not show any pull-down by His 6 -NSP2, supporting the specificity of the NSP2-GIGYF2 interaction. We also successfully detected this interaction when both His 6 -NSP2 and GST-GIGYF2 743À1,085 were co-expressed in E. coli. In this case, a GST pull-down assay was performed using an E. coli lysate and a specific His 6 -NSP2 retention on GST-GIGYF2 743À1,085 -bound beads was detected by both Coomassie blue staining and Western blot ( Figure S2C). Similarly, we sought to test the capacity of His 6 -NSP2 to bind the N-terminal region of GIGYF2 (residues: 1-267), but failed to obtain this recombinant GIGYF2 fragment with a sufficient yield and solubility (data not shown).
Our co-IP experiments pointed out the 1-350 region of NSP2 as the minimal segment required to bind GIGYF2 in cellulo ( Figure 1E). In attempt to narrow down this region, we generated three recombinant ll OPEN ACCESS 4 iScience 25, 104646, July 15, 2022 iScience Article His 6 -tagged fragments of NSP2 covering the 1-350 region (residues: 1-115, 107-212, and 204-350) to test their direct interaction with GIGYF2 743À1,085 . Using a His pull-down assay, we found that none of these fragments successfully retained GIGYF2 743À1,085 , with the exception of full-length His 6 -NSP2 ( Figure S2D), confirming a large interaction surface between NSP2 and GIGYF2. Because an interaction was detected between NSP2 and the 4EHP-binding region of GIGYF2 (residues: 1-267) by co-IP, we also speculated that 4EHP could bind NSP2 independently of GIGYF2. To test this, we used a V5-tagged version of 4EHP carrying the W95A substitution which is known to disrupt its interaction with GIGYF2 (Peter et al., 2017). HEK293T cells were co-transfected by vectors encoding Flag-NSP2 and V5-4EHP W95A , or its WT counterpart. Using Flag IPs, we found that both WT and W95A versions of 4EHP were co-immunoprecipitated by NSP2 to a comparable extent ( Figure 2D), indicating that 4EHP can bind NSP2 independently of GIGYF2. This result prompted us to evaluate their direct association in vitro. As performed with GIGYF2 743À1,085 , full-length GST-fused 4EHP was expressed in E. coli and the purified untagged protein was incubated with His 6 -NSP2 to perform a His pull-down assay. In agreement with our co-IP results, we observed a specific binding of 4EHP to His 6 -NSP2, thus confirming their interaction in a GIGYF2-independent manner ( Figure S2E). Conversely, we found that NSP2 can simultaneously bind both 4EHP and GIGYF2 743À1,085 since the latter two were retained in our His 6 -NSP2 pull-down when all three recombinant proteins were incubated at the same time ( Figure 2E). Gupta et al. recently reported that a natural variant of NSP2 carrying the G262V/G265V substitutions showed a reduced interaction with 4EHP-GIGYF2 by affinity purification-mass spectrometry (Gupta et al., 2021). We therefore tested whether these G to V substitutions could impact the integrity of the NSP2/4EHP-GIGYF2 complex by co-IP and in vitro interaction assays. Our co-IP experiment with a iScience Article Flag-tagged NSP2 harboring the G262V/G265V variation confirmed that substituting G262 and G265 with valines reduced its interaction with endogenous GIGYF2, 4EHP, and ZNF598 ( Figure S3A). By contrast, when introduced in the His 6 -NSP2 recombinant protein, these two substitutions did not change the capacity of NSP2 to retain both recombinant 4EHP and GIGYF2 743À1,085 in a His pull-down assay (Figure S3B). Nevertheless, it is worth noting that the direct contribution of the G262 and G265 residues of SARS-CoV-2 NSP2 in binding 4EHP-GIGYF2 remains unclear since both are not found in SARS-CoV-1 and MERS-CoV while the NSP2/4EHP-GIGYF2 interaction is detected across these two b-CoVs ( Figure S3C). Overall, these data suggest that the mode of interaction between NSP2 and the intact 4EHP-GIGYF2 complex involves multiple interaction interfaces, and may be more sophisticated in cellulo than in our in vitro binding assays.

NSP2 reduces the silencing capacity of GIGYF2
The role of 4EHP-GIGYF2 as a translational repressor has been described in various contexts (for review, see (Christie and Igreja, 2021)). Having shown that NSP2 directly targets the 4EHP-GIGYF2 complex through at least two contact points, namely the region 743-1,085 of GIGYF2 and 4EHP itself, we wished to assess whether the silencing capacity of 4EHP-GIGYF2 could be altered by NSP2. To address this point, we used the lN-BoxB tethering approach in HEK293T cells (Baron-Benhamou et al., 2004;Pillai et al., 2004). A Renilla luciferase reporter mRNA containing five BoxB sequences (RLuc-5BoxB) in the 3 0 untranslated region (3 0 UTR) was co-expressed with a plasmid expressing V5-tagged GIGYF2 fused to a lN peptide, which has a high affinity for the BoxB sequences ( Figure 3A). The silencing capacity of GIGYF2 was assessed following the measurement of luciferase activity in HEK293T expressing Flag-NSP2 or Flag as control.
The expression of Flag-NSP2 and lN-GIGYF2 was also verified by WB ( Figure 3C). As expected, we observed that the recruitment of GIGYF2 to the 3 0 UTR markedly reduced luciferase activity in control cells (5-fold repression). Interestingly, this lN-GIGYF2-mediated repression was reduced upon expression of Flag-NSP2 (3-fold repression), indicating an inhibitory effect of NSP2 on GIGYF2 function in silencing ( Figure 3A).
Several studies have reported that GIGYF2 has two distinct mechanisms of repression: one is 4EHP-dependent and affects translation; the other is 4EHP-independent and involves the deadenylase activity of the CCR4-NOT complex (Amaya Ramirez et al., 2018). Our co-IP data showed that NSP2 did not interact with CCR4-NOT ( Figure 1A), suggesting that it should preferentially impair the 4EHP-dependent activity of GIGYF2. To test this, lN-GIGYF2 was tethered to a RLuc-5BoxB mRNA containing a self-cleaving hammerhead ribozyme (HhR) at the 3 0 -end to generate a poly(A) stretch of 114 nucleotides, followed by 40 nucleotides to block CCR4-NOT-dependent deadenylation (RLuc-5BoxB-A 114 -N 40 -HhR), as previously described . Tethering GIGYF2 to RLuc-5BoxB-A 114 -N 40 -HhR induced a 2.2-fold repression of this reporter in cells transfected with a control vector ( Figure 3B). Owing to its inability to encounter mRNA deadenylation, the silencing magnitude of this reporter remained $40% lesser than the one induced when GIGYF2 is tethered to RLuc-5BoxB. By contrast, we observed that NSP2 expression decreased the GIGYF2-mediated repression of the reporter to 1.4-fold ( Figure 3B), indicating a specific role of NSP2 in blocking the deadenylation-independent silencing capacity of GIGYF2.

NSP2 impairs the contribution of 4EHP-GIGYF2 into miRNA-mediated silencing
The activity of the 4EHP-GIGYF2 complex is required for the optimal translation repression driven by several pathways in a large range of cellular processes (Christie and Igreja, 2021). In particular, the miRNA-induced translational repression is affected by the 4EHP-GIGYF2 complex in human cells (Chen and Gao, 2017;Chapat et al., 2017b;Schopp et al., 2017). To evaluate the impact of NSP2 in miRNA-mediated silencing, we transiently transfected HEK293T cells with a Renilla luciferase construct either lacking (RLuc), or containing six bulged let7a miRNA-binding sites in its 3 0 UTR (RLuc-6let7a), together with a firefly luciferase construct (FLuc) as a transfection control ( Figure 4A). Normalized RLuc activity was markedly reduced by the presence of let7a-binding sites, with $40-fold repression in cells expressing a control vector. By contrast, expression of NSP2 resulted in a decreased repression magnitude of the RLuc-6let7a reporter compared to RLuc ($30-fold), indicating a reduced let7a-mediated silencing in the presence of NSP2 ( Figure 4A).
Evidence suggests that the 4EHP-GIGYF2 complex contributes to miRNA action through the binding of GIGYF2 to a proline-rich motif (PPGL) in the silencing domain (SD) of GW182, the scaffolding protein of miRISC ( iScience Article silencing domain of GW182 was artificially tethered to the RLuc-5BoxB reporter in control and Flag-NSP2expressing cells. We found that tethering GW182 SD induced a $40-fold repression of the RLuc-5BoxB level in the control cells, while this silencing effect was reduced in NSP2-expressing cells ($30-fold; Figure 4B). As a control, a GW182 SD mutant carrying a deletion of the GIGYF2-binding motif (DPPGL) was used to evaluate the contribution of GIGYF2. In control cells, tethering the DPPGL mutant of GW182 SD engendered a silencing of RLuc-5BoxB that is $35% lesser than the one induced by tethering WT GW182 SD , confirming the importance of the GIGYF2/PPGL interaction for the GW182-mediated silencing. Tethering this DPPGL iScience Article mutant in cells expressing Flag-NSP2 showed that its silencing activity is not affected by NSP2, indicating that NSP2 action on GW182 SD -mediated silencing is exerted through targeting GIGYF2 ( Figure 4B).
Tethering GW182 SD to an mRNA is known to lead to translational repression, deadenylation, and degradation of the target mRNA, while 4EHP-GIGYF2 was only shown to participate to translational repression iScience Article (Jonas and Izaurralde, 2015;Schopp et al., 2017). To uncouple 4EHP-GIGYF2-driven translational repression from mRNA destabilization effects, GW182 SD was tethered to the RLuc-5BoxB-A 114 -N 40 -HhR reporter in control and Flag-NSP2-expressing cells. In this case, the GW182 SD -mediated repression was reduced to $12-fold in control cells due to the inability of RLuc-5BoxB-A 114 -N 40 -HhR to encounter mRNA deadenylation, and to $8-fold when the DPPGL mutant was used ( Figure 4C). Upon NSP2 expression, the silencing activity of GW182 SD was decreased to $8-fold but remains unchanged with its DPPGL version, confirming a specific alteration of the 4EHP-GIGYF2-driven translational repression by NSP2. Altogether, these results indicate that the GIGYF2-dependent repressive activity of GW182 SD is partially altered upon expression of NSP2 in human cells.

DISCUSSION
Upon infection, SARS-CoV-2 impairs splicing, export, translation, and degradation of host mRNAs (Finkel et al., 2021;Banerjee et al., 2020). Here, we present evidence to support a new layer of complexity in the post-transcriptional alteration of the host transcriptome by SARS-CoV-2. We propose that SARS-CoV-2 NSP2 directly targets the 4EHP-GIGYF2 complex to decrease its silencing capacity ( Figure 4D). While our model lacks the context of other viral proteins that would be present in a bona fide infection, this mechanism could nonetheless unveil the impact of NSP2 on the post-transcriptional silencing of gene expression of human cells, pointing out 4EHP-GIGYF2 targeting as a possible strategy of SARS-CoV-2 to take over the silencing machinery and to suppress host defenses. Further studies in a more physiological context, such as lung/airway cell lines or SARS-CoV-2-infected samples, should help resolve this conundrum.
How does NSP2 impair 4EHP-GIGYF2 function? Combining co-IP experiments and in vitro binding assays with recombinant proteins, we concluded that NSP2 uses its N-terminal region encompassing its conserved zinc finger domain, to interact with the 4EHP-GIGYF2 complex. Our pull-down assays indicate the direct interaction of NSP2 with both 4EHP and two domains from GIGYF2, confirming a sophisticated mode of binding in cellulo. While we searched for the minimal region of NSP2 required for these interactions, we failed to narrow down a fragment smaller than the 1-350 region since truncations at both extremities of this domain abrogate its binding to 4EHP-GIGYF2 ( Figures 1E and S2D). Nevertheless, this remains in agreement with Gupta et al. who pointed out that the G262V and G265V mutations located within this region of NSP2 reduced binding to 4EHP-GIGYF2 (Gupta et al., 2021). This natural variation occurs in a poorly conserved patch in NSP2 that is subsequently becoming more hydrophobic due to the G to V substitution. It is worth noting that G262 and G265 are not conserved across the SARS-CoV-1 and MERS-CoV ( Figure S3C), while the NSP2/4EHP-GIGYF2 interaction exists among these viruses. The direct contribution of the G262/G265 residues of NSP2 in binding 4EHP-GIGYF2 is therefore questionable. It is tempting to speculate that the G to V variation could rather increase NSP2's affinity for host interactors that outcompete 4EHP-GIGYF2. This point is supported by the affinity purification-mass spectrometry made by Gupta et al., showing that the G262V/G265V variation increases the affinity of NSP2 for factors such as the mitochondrial protein UQCRC1, or the actin-nucleation-promoting protein WASHC5 (Gupta et al., 2021). This could explain why the G262V/G265V variation reduces the NSP2/4EHP-GIGYF2 interaction in cellulo, but not in our in vitro pull-down assay lacking the context of other host interactors of NSP2. Further investigation will be needed to fully resolve the structural basis of the NSP2/4EHP-GIGYF2 complex and thus elucidate NSP2 action on translation silencing. In particular, it now remains to be determined whether NSP2 binding induces conformational changes in 4EHP-GIGYF2 that impair either the cap-binding pocket of 4EHP, or influence the recruitment of GIGYF2's co-factors such as CCR4-NOT and DDX6.
Global measurement of miRNA action showed that translational repression accounts for 6%-26% of the silencing of each mRNA target in mammalian cells, and 4EHP-GIGYF2-mediated translational repression is observed at early time points of the silencing process (Schopp et al., 2017;Eichhorn et al., 2014). Consistent with these observations, the impact of NSP2 remains mild on the let7a-targeted RLuc reporter (Figure 4A). The latter is also known to underestimate the contribution of the translational repression to the silencing process since mRNA destabilization is the dominant effect of miRNA-mediated silencing at steady state (Eichhorn et al., 2014;Bethune et al., 2012). Our tethering assays with GW182 SD have proven helpful to overcome this limitation. The derepression of the GW182 SD -induced silencing of RLuc-5BoxB-A 114 -N 40 -HhR upon deletion of the PPGL motif supports the fact that our tethering assay faithfully recapitulates the contribution of GIGYF2 into miRNA-induced translation repression. Consistently, these tethering assays showed that the silencing capacity of GW182 SD upon NSP2 expression equals the one of its ll OPEN ACCESS iScience 25, 104646, July 15, 2022 9 iScience Article DPPGL version in control cells (Figures 4B and 4C), indicating that the contribution of 4EHP-GIGYF2 into GW182 SD -mediated translation repression is fully targeted by NSP2. Indeed, this impact of NSP2 will need to be investigated in further more physiological studies using endogenous miRNAs and transcripts.
At the moment, it is uncertain what the functional interplay between SARS-CoV-2 infection and miRNAmediated silencing is in human cells. Host miRNAs are known to be produced as a part of antiviral response to counteract the infection by targeting viral transcripts, although SARS-CoV-2 infection was recently shown to have minimal impact on the miRNA repertoire of its host cell (Cullen, 2006;Bruscella et al., 2017;Pawlica et al., 2021). Computational analyses have predicted the presence of many putative miRNA-binding sites on the SARS-CoV-2 genome, suggesting that the SARS-CoV-2 genome could be actively targeted by host miRNAs (Xie et al., 2021;Arisan et al., 2020;Chow and Salmena, 2020). Particularly worth mentioning is the work of Xie et al. which recently identified let-7-binding sites in the coding sequence of S and M proteins of SARS-CoV-2 genome, and experimentally confirmed that let-7 blocks SARS-CoV-2 replication by targeting S and M proteins (Xie et al., 2021). Through the NSP2/4EHP-GIGYF2 axis, SARS-CoV-2 could therefore escape from the host defense system by impairing the function of the effector machinery of miRNAs. The recent discovery that 4EHP and GIGYF2 are needed for infection by SARS-CoV-2 could reinforce this idea, although further research will be required to test this possibility (Hoffmann et al., 2021).
While the silencing capacity of miRISC is partially impeded upon NSP2 expression, there is no guarantee that miRNA action is the prime target of NSP2. The activity of 4EHP-GIGYF2 is mobilized by several pathways, one of which may be more affected than miRNAs. These pathways include TTP-and ZNF598-mediated mRNA silencing, as well as the repression of mRNAs with altered ribosome activity or premature termination codons as part of the nonsense-mediated mRNA decay pathway (Christie and Igreja, 2021). Future studies are thus mandatory to evaluate the potential impact of NSP2 in modulating these processes in human cells. In the case of miRNA, the alteration of let-7a-mediated inhibition by NSP2 could be extrapolated to other miRNAs whose action relies on 4EHP, such as miR-145 or miR-34a (Jafarnejad et al., 2018;Zhang et al., 2021). Recent evidence demonstrated that the 4EHP/miR-34a axis is required for the translational repression of mRNAs encoding IFN-b through targeting the 3 0 UTR of Ifnb1 mRNA . Beyond miRNA, 4EHP-GIGYF2 also controls the production of TTP-targeted mRNAs that encode inflammatory cytokines such as TNF-a and IL-8 (Fu et al., 2016;Tollenaere et al., 2019). In this context, a possible consequence of NSP2 expression could be the overproduction of early response pro-inflammatory cytokines. Exploring this point would be of utmost importance since impaired type I interferon activity and inflammatory responses are detected in patients with severe COVID-19 (Hadjadj et al., 2020). To examine whether NSP2 could impact the function of 4EHP in regulating IFN-b expression, we expressed NSP2 along with a reporter construct containing the 3 0 UTR of Ifnb1 mRNA into HEK293T cells ( Figure S4A). Remarkably, the reporter expression was repressed $2.9-fold in control cells, but only $1.6-fold in NSP2expressing cells, indicating that NSP2 could potentially unbalance the production of IFN-b through the Ifnb1 3 0 UTR ( Figure S4B). With this in mind, further investigations into whether the NSP2/4EHP-GIGYF2 axis can dysregulate sustained cytokine production may therefore prove useful.
In conclusion, our study raises the possibility that SARS-CoV-2 could target the human 4EHP-GIGYF2 complex to selectively modulate its capacity to effect translation repression. Our model may represent a novel framework to investigate the mechanisms underlying the pathogenicity of SARS-CoV-2 based on the interaction of NSP2 with 4EHP-GIGYF2. Ultimately, we hope that this study will be a primer for further more physiological research to evaluate the generalizability of our model.

Limitations of the study
Although our work depicts how the 4EHP-GIGYF2 complex is compromised by the SARS-CoV-2 protein NSP2, it is not without limitations. The molecular mechanism described in this article comes from in vitro studies using the HEK293T cells, which were chosen to dissect the NSP2/4EHP-GIGYF2 interaction on the basis of their capacity to express both NSP2 and 4EHP-GIGYF2 at a high level. However, our model will need to be validated in a more physiological context, such as lung/airway cell lines or SARS-CoV-2-infected samples, to assess the functional relevance of this interaction. In this sense, whether other SARS-CoV-2 proteins interact with NSP2 and thereby prevent its interaction with GIGYF2/4EHP cannot be dismissed. Our conclusion that NSP2 impairs miRNA action is mainly based on the use of artificial reporters (miRNA and tethering assays). To date, these assays have proven invaluable in dissecting the functions ll OPEN ACCESS iScience 25, 104646, July 15, 2022 iScience Article of numerous silencing factors in human cells. Nevertheless, our data will need to be complemented by follow-up studies to investigate the impact of NSP2 on endogenous transcripts. Future investigations that employ transcriptomic analysis of the NSP2-mediated post-transcriptional alterations will undoubtedly extend our view on the repertoire of endogenous miRNA/mRNA pairs which are affected by NSP2.
Another potential limitation of our study is that we did not explore potential mechanisms for NSP2-mediated inhibition of 4EHP-GIGYF2 and how they could cause the pathogenicity of SARS-CoV-2. Future studies will be needed to dissect the molecular basis of this mechanism using additional models of coronavirus infection.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Clé ment Chapat (clement.chapat@cnrs.fr).

Materials availability
All newly created cell populations generated in this study are available upon request.
Data and code availability d All data reported in this paper will be shared by the lead contact upon request. d This paper does not report original code. d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Cell lines and culture
HEK293T cells (Sigma-Aldrich) were routinely maintained in high glucose Dulbecco's Modified Eagle's Medium (DMEM) with GlutaMAX supplemented with 10% fetal bovine serum (FBS) and 2% penicillin/streptomycin in a humidified atmosphere of 5% CO2 at 37 C. Flp-In T-REx 293 cells (Thermo Fisher Scientific) were grown in similar conditions supplemented with 100 mg/mL zeocin and 15 mg/mL blasticidin. The absence of mycoplasma contamination in cells was routinely tested. The cell line inducibly expressing Flag-NSP2 was generated by co-transfecting pcDNA5-FRT-TO-FH-Nsp2 (Addgene plasmid 157683) and pOG44 (Thermo Fisher Scientific) with a 1:10 ratio. Transfected cells were selected and maintained in media supplemented with 100 mg/mL hygromycin. Expression of Flag-tagged proteins was induced for 24 h by addition of tetracycline to 1 mg/mL final concentration.

METHOD DETAILS
CRISPR/cas9-mediated genome editing CRISPR-Cas9-mediated genome editing of HEK293 cells was performed according to Ran et al. The following oligonucleotides encoding a small guide RNA cognate to the coding region of Gigyf2 gene were used: 5 0 -CACCGGGAGGAACCCCTTCCACCAT and 5 0 -AAACATGGTGGAAGGGGTTCCTCCC. These oligos which contain BbsI restriction sites were annealed creating overhangs for cloning of the guide sequence oligos into pSpCas9(BB)-2A-Puro (PX459) V2.0 (Addgene plasmid 62988) by BbsI digestion. To generate KO HEK293 cells, we transfected 700,000 cells with the pSpCas9(BB)-2A-Puro plasmid. 24 h after transfection, puromycin was added in the cell medium to 1.5 mg/mL final concentration. After 72 h, puromycin-resistant cells were isolated into 96-well plates to obtain monoclonal colonies. Clonal cell populations were analyzed by WB for protein depletion.

Plasmid cloning
For Flag-NSP2 IP, pcDNA5-FRT-TO-FH-Nsp2 (Addgene plasmid 157683) was used. Truncated versions of NSP2 were generated by replacing the sequence encoding full-length NSP2 by PCR-amplified fragments at the BamHI-XhoI sites of the pcDNA5-FRT-TO-FH vector. The full-length cDNAs encompassing the coding region of human 4EHP and GIGYF2 were obtained by RT-PCR using total RNA from HEK293T cells and cloned into the pCI-Neo vector (Promega) at the XhoI-NotI sites in frame with a sequence encoding a V5 tag inserted at the NheI-XhoI sites. To generate the pCI-lNV5-GIGYF2 vector, a fragment containing the GIGYF2 sequence was obtained by PCR and inserted into the pCI-lNV5 at the XhoI-NotI sites. A similar strategy was used to generate the V5-tagged fragments of GIGYF2 using primers to isolate the following domains encompassing residues: 1-267; 258-495; 486-752; 743-1,085; 1,076-1,320. To construct the pCI-lNV5-GW182 SD vector, a fragment encoding the residues 1382-1690 was cloned by PCR as a XhoI-NotI fragment from the pFRT/TO/FLAG/HA-DEST TNRC6C vector (Addgene plasmid 19885) (Landthaler et al., 2008) and inserted into the pCI-lNV5 at the XhoI-NotI sites. For recombinant GST-fused protein expression, full-length 4EHP or GIGYF2 743À1,085 coding sequences contained in PCR-amplified BamHI-NotI or XhoI-NotI fragments, respectively, were cloned into a pGEX-6P-1 vector (Amersham) in frame with the GST coding sequence. The pET-28b vector (EMD Biosciences) was used to express NSP2 as a ll OPEN ACCESS iScience Article