key: cord-271970-i35pic5o
authors: Boris, Bonaventure; Antoine, Rebendenne; de Gracia Francisco, Garcia; Marine, Tauziet; Joe, McKellar; Valadão Ana Luiza, Chaves; Valérie, Courgnaud; Eric, Bernard; Laurence, Briant; Nathalie, Gros; Wassila, Djilli; Mary, Arnaud-Arnould; Hugues, Parrinello; Stéphanie, Rialle; Olivier, Moncorgé; Caroline, Goujon
title: A genome-wide CRISPR/Cas9 knock-out screen identifies the DEAD box RNA helicase DDX42 as a broad antiviral inhibitor
date: 2020-10-28
journal: bioRxiv
DOI: 10.1101/2020.10.28.359356
sha: 
doc_id: 271970
cord_uid: i35pic5o

Genome-wide CRISPR/Cas9 knock-out genetic screens are powerful approaches to unravel new regulators of viral infections. With the aim of identifying new cellular inhibitors of HIV-1, we have developed a strategy in which we took advantage of the ability of type 1 interferon (IFN) to potently inhibit HIV-1 infection, in order to create a cellular environment hostile to viral replication. This approach led to the identification of the DEAD-box RNA helicase DDX42 as an intrinsic inhibitor of HIV-1. Depletion of endogenous DDX42 using siRNA or CRISPR/Cas9 knock-out increased HIV-1 infection, both in model cell lines and in physiological targets of HIV-1, primary CD4+ T cells and monocyte-derived macrophages (MDMs), and irrespectively of the IFN treatment. Similarly, the overexpression of a dominant-negative mutant of DDX42 positively impacted HIV-1 infection, whereas wild-type DDX42 overexpression potently inhibited HIV-1 infection. The positive impact of endogenous DDX42 depletion on HIV-1 infection was directly correlated to an increase in viral DNA accumulation. Interestingly, proximity ligation assays showed that DDX42, which can be mainly found in the nucleus but is also present in the cytoplasm, was in the close vicinity of HIV-1 Capsid during infection of primary monocyte-derived macrophages. Moreover, we show that DDX42 is also able to substantially decrease infection with other retroviruses and retrotransposition of long interspersed elements-1 (LINE-1). Finally, we reveal that DDX42 potently inhibits other pathogenic viruses, including Chikungunya virus and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2).

Over the past 20 years, a growing list of cellular proteins with various functions have been identified as capable of limiting different steps of HIV-1 life cycle (Doyle et al., 2015; Ghimire et al., 2018) . Lentiviruses have generally evolved to counteract the action of these so-called restriction factors. However, type 1 interferons (IFNs) induce, through the expression of interferonstimulated genes, an antiviral state particularly efficient at inhibiting HIV-1 when cells are preexposed to IFN (Ho et al., 1985; Bednarik et al., 1989; Coccia et al., 1994; Baca-Regen et al., 1994; Goujon and Malim, 2010; Cheney and McKnight, 2010) . The dynamin-like GTPase MX2, and, very recently, the restriction factor TRIM5a, have both been shown to participate in this IFNinduced inhibition (Goujon et al., 2013a; Kane et al., 2013; OhAinle et al., 2018; Jimenez-Guardeño et al., 2019) . With the hypothesis that additional HIV-1 inhibitors remained to be identified, we took advantage of the hostile environment induced by IFN to develop a wholegenome screen strategy in order to reveal such inhibitors.

The development of CRISPR/Cas9 as a genome editing tool in mammalian cells has been a major breakthrough, notably with the generation of pooled single guide (sg)RNA libraries delivered with lentiviral vectors (LVs), allowing high-throughput screens at the whole-genome scale (Shalem et al., , 2015 Doench, 2018) . We used the Genome-Scale CRISPR Knock-Out (GeCKO) sgRNA library developed by Feng Zhang's laboratory Shalem et al., 2014 Shalem et al., , 2015 to generate cell populations knocked-out for almost every human gene in the T98G glioblastoma cell line. This model cell line is both highly permissive to lentiviral infection and potently able to suppress HIV-1 infection following IFN treatment (Supplementary Information, SI Figure 1 ). The screen strategy is depicted in Figure 1A . T98G cells were first modified to stably express Cas9 and a high number of T98G-Cas9 cells were then transduced with LVs coding sublibraries A or B, the two halves of the GeCKO library. A low multiplicity of infection (MOI) was used to avoid multiple integration events and increase the probability to express only 1 sgRNA per cell. Deep sequencing analysis of the GeCKO cell populations showed more than 94% coverage for both libraries (≥ 10 reads for 61,598 and 54,609 sgRNA-coding sequences out of GeCKO populations were subjected to type 1 IFN treatment in order to induce the antiviral state and, 24h later, incubated with VSV-G-pseudotyped, HIV-1 based LVs coding for an antibiotic resistance cassette. Two days later, the cells successfully infected despite the IFN treatment were selected by cell survival in the presence of the corresponding antibiotic. In order to enrich the population with mutants of interest and to limit the presence of false-positives, two additional rounds of IFN treatment, infection and selection (using different antibiotics) were performed ( Figure 1A ). As expected, the cells enriched after each round of the screen became less refractory to HIV-1 infection following IFN treatment (SI Figure 2) . . The GeCKO populations were then exposed to IFN for 24h and challenged with HIV-1-based LVs coding for an antibiotic resistance gene. After selection by antibiotic addition, the surviving cells (i.e. efficiently infected despite the IFN treatment) were amplified. In total, the GeCKO population underwent three successive rounds of IFN treatment, infection and selection using LVs coding for different resistance cassettes. The genomic DNAs of the initial GeCKO populations and the three-time selected populations were extracted, the sgRNAcoding sequences were amplified by PCR and sequenced by next generation sequencing (NGS). B. The candidate genes were identified using the MAGeCK computational statistical tool (Li et al., 2014) . MAGeCK establishes a Robust Rank Aggregation (RRA) score for each gene based on the sgRNA enrichment and the CTRL  IFNAR1  MX2  WARS2  DDX42  CLDN12  REEP1  IKBIP  TOPBP1  COPG1  FGF4  SPRYD7  RDH8  DCTPP1  PRLR  SUN3  CXorf27  TMEM161A  KCN6  LSM5  LOC339862  PKD2  NDPC1  KIAA1549  mir-4737  VSTM2A  CALU  SMARCA2  DNAJB2  FAM212A  ETNPPL  RHOC  mir-105-1  CXorf56  CCSER1  SLC25A2  OR6N2  UBA1  HSPA1B  SIAE  KRT222  ADRA1A  LIN54  TECTB  CDRT1  TCEAL7  TRERF1  CD14  HSP90AA1  CTNNA2  PLAC1L  RELB  BSND   0 

GeCKO population versus Selected population

number of sgRNAs targeting the same gene. Here, genes belonging to the IFN-response pathway (indicated in blue) and DDX42 (in red) are represented (together with their respective rank into brackets) for the 2 independent screens (the results of which were merged in the analysis). C. T98G/CD4/CXCR4/Cas9/Firefly KO populations were generated for the 25 best candidate genes of each screen. The control (CTRL) condition represents the mean of four negative control cell populations (i.e. expressing 4 different non-targeting sgRNAs) and IFNAR1 and MX2 KO cell populations were used as positive controls. KO cell populations were pre-treated with IFN and infected with HIV-1 Renilla. The cells were lysed 24 h post-infection and the two luciferase signals were measured (Renilla signals were normalized to internal Firefly control). The IFN inhibition (corresponding to the ratio of the untreated / IFN-treated conditions) was calculated and sets at 100% inhibition for the average of the 4 negative CTRL populations. A representative experiment is shown (mean and standard deviation from technical duplicates).

To identify the genes of interest, the differential sgRNA abundance between the starting GeCKO populations and the enriched (3-times selected) populations was analysed by NGS. 2,332 and 3,900 different sgRNAs were identified (≥ 10 reads) for screens A and B, which represented 3,6% and 7% of the sgRNAs present in the initial GeCKO population A and B, respectively. The MAGeCK algorithm, which assigns a robust ranking aggregation (RRA) score, was used to rank the gene candidates from each screen ( Figure 1B ). For both screens, we observed a positive enrichment for 200 genes (RRA score > 0,01), with the best hits being IFNAR1, JAK1 and STAT2 ( Figure 1B ). All the crucial mediators of the type 1 IFN signalling cascade were present among the top hits in both screens (with the notable exception of STAT1), validating our approach and confirming the identification of relevant genes. Interestingly, most of the other 195 positively selected genes displayed unknown functions or functions that were a priori unrelated to the IFN response pathway or to innate immunity. Of note, very little overlap was observed between the two independent screens, performed with two different sub-libraries. However, a poor overlap between independent screens has been observed before and does not preclude obtaining valid data (Doench, 2018) . Therefore, the top 25 candidate genes from each independent screen were selected for further validation. As a first validation step, the sequences of the most enriched sgRNA for each gene were chosen and cloned into the LentiGuide-Puro vector . T98G-Cas9 cells expressing HIV-1 CD4 and CXCR4 receptors, as well as the firefly luciferase as internal control (T98G Cas9/CD4/CXCR4/Firefly cells), were transduced with the sgRNA-expressing LVs to generate individual KO populations. Four irrelevant, non-targeting sgRNAs, as well as sgRNAs targeting IFNAR1 and MX2, were used to generate negative and positive control populations, respectively. The KO cell populations were pre-treated with IFN and infected with an HIV-1 reporter virus expressing the Renilla luciferase reporter and bearing HIV-1 envelope (NL4-3/Nef-IRES-Renilla, hereafter called HIV-1 Renilla). Infection efficiency was analysed 30h later ( Figure 1C ). As expected, IFNAR1 and MX2 KO fully and partially rescued HIV-1 infection from the protective effect of IFN, respectively (Goujon et al., 2013a; Bulli et al., 2016; Xu et al., 2018) . The KO of two candidate genes, namely WARS2 and DDX42, allowed a partial rescue of HIV-1 infection from the IFN-induced inhibition, suggesting a potential role of these candidate genes.

DDX42 is a member of the DExD/H box family of RNA helicases with RNA chaperone activities (Uhlmann-Schiffler et al., 2006) and, as such, retained our attention. Indeed, various DEAD box helicases, such as DDX3, DDX6 and DDX17, are well-known to regulate HIV-1 life cycle (Gringhuis et al., 2017; Sithole et al., 2018 Sithole et al., , 2020 Soto-Rifo et al., 2013; Williams et al., 2015; Yedavalli et al., 2004) . However, to our knowledge, the impact of DDX42 on HIV-1 replication had never been studied. In order to validate the effect of DDX42 KO on HIV-1 infection in another model cell line, two additional sgRNAs were designed (sgRNA-2 and -3) and used in parallel to the one identified in the GeCKO screen (sgDDX42-1) (Figure 2A ). U87-MG/CD4/CXCR4 cells were used here, as we previously extensively characterized the IFN phenotype in these cells (Goujon et al., 2013a) . Control and DDX42 KO cell populations were treated or not with IFN for 24h prior to infection with increasing amounts of HIV-1 Renilla. DDX42 depletion improved HIV-1 infection with all three sgRNAs used, confirming that endogenous DDX42 had a negative impact on HIV-1 infection. Interestingly, the increase in infection efficiency induced by DDX42 KO was observed irrespectively of the IFN treatment. DDX42 is not known to be an ISG (interferome database and our previous study (Goujon et al., 2013a) , GEO accession number: GSE46599), which we confirmed in a number of cell types (SI Figure 3 ). The fact that the IFN-induced state is at least partially saturable (SI Figure 1 ) explains why an intrinsic inhibitor of HIV-1, which is not regulated by IFN, could be identified by our approach: removing one barrier to infection presumably rendered the cells generally more permissive and, in this context, IFN had less of an impact. sgDDX42-2, and sgDDX42-3) and 4 different non-targeting sgRNAs, respectively (for the CTRL condition, the average of the data obtained with the four cell populations is shown). Cells were pre-treated or not with IFN 24 h prior to infection with HIV-1 Renilla (50 ng p24 Gag ) and the ratio of Renilla/firefly activity was analysed, as in 3) and the infection efficiency was measured by p24 Gag intracellular staining followed by flow cytometry analysis.

When indicated, the cells were treated with 10 µM zidovudine (azidothymidine, AZT) and lamivudine (3TC) reverse transcription inhibitors for 2h prior to infection. D. Blood monocytes from healthy donors were isolated, differentiated into MDMs, and transfected with non-targeting siRNAs (siCTRL1 and siCTRL2) or siRNAs targeting DDX42 (siDDX42-1 and siDDX42-2). Two days after transfection, MDMs were infected with a CCR5tropic version of NL4-3-Renilla (100 ng p24 Gag ). Infection efficiencies were monitored 24h later by measuring Renilla activity. The relative luminescence results from experiments performed with cells from 3 different donors are shown. E. CD4+ T cells were isolated from peripheral blood mononuclear cells, activated with IL-2 and phytohemagglutinin, and electroporated with Cas9-sgRNA RNP complexes using two non-targeting sgRNAs (sgCTRL1 and sgCTRL2) and five sgRNAs targeting DDX42 (sgDDX42-1, -2, -3, -4, -5) two days later. Four days after electroporation, the activated CD4+ T cells were infected with NL4-3 Renilla for 24h. Relative infection efficiencies obtained with cells from three independent donors are shown. DDX42 protein levels were determined by immunoblot and Actin served as a loading control (a representative experiment is shown In order to confirm DDX42's effect on HIV-1 infection with an independent approach, we used 3 different siRNAs to knockdown DDX42 expression. We observed that depleting DDX42 with siRNAs (with >90% efficiency both at the mRNA and protein levels, Figure 2B ) improved HIV-1 infection efficiency by 3 to 8-fold when using an HIV-1 Renilla reporter in U87-MG/CD4/CXCR4 cells, irrespectively of the presence of IFN ( Figure 2B , right panel). Of note, wild-type HIV-1 infection was also impacted by DDX42 silencing, as shown by Capsid (p24 Gag ) intracellular staining 30h post-infection ( Figure 2C ). We then investigated whether DDX42 had an impact in HIV-1 primary target cells. In MDMs, we observed that HIV-1 infection was increased by about 2fold following DDX42 silencing ( Figure 2D ), whereas DDX42 mRNA abundance was decreased by only 40% in these cells using siRNAs (SI Figure 4 ). As the siRNA approach did not work in our hands in primary T cells, we used electroporation of pre-assembled Cas9-sgRNA ribonucleoprotein complexes (RNPs) to deplete DDX42 in primary CD4+ T cells ( Figure 2E ).

Highly efficient depletion of DDX42 was obtained with all 5 sgRNAs as compared to the 2 sgCTRLs ( Figure 2E , bottom panel) and this depletion increased HIV-1 infection by 2-to 3-fold,

showing a role of DDX42 as an intrinsic inhibitor of HIV-1 in primary CD4+ T cells.

Having established that endogenous DDX42 had an impact on HIV-1 infection, we then analysed the consequences of DDX42 overexpression. An irrelevant control (Firefly) or DDX42 were ectopically expressed in U87-MG/CD4/CXCR4 and the cells were challenged with HIV-1 Renilla ( Figure 2F ). DDX42 ectopic expression induced a substantial inhibition of HIV-1 infection (about 5-fold decrease in infection efficiency in comparison to the control) ( Figure 2F ). We then tested a mutant version of DDX42 that is unable to hydrolyse ATP and may supposedly act as a dominant negative, DDX42 K303E (Granneman et al., 2006; Rocak, 2005 ) (SI Figure 5) . Interestingly, the expression of DDX42 K303E mutant increased HIV-1 infection by 3-fold, reminiscent of what we observed with DDX42 depletion. Altogether, these data showed for the first time that endogenous DDX42 is able to intrinsically inhibit HIV-1 infection.

In order to determine the step of HIV-1 life cycle affected by DDX42, we first analysed viral entry with a BlaM-Vpr assay (Cavrois et al., 2002) . Consistent with the observation that VSV-Gpseudotyping did not bypass DDX42-mediated inhibition of infection (SI Figure 6 ), we observed that DDX42 silencing did not impact HIV-1 entry (SI Figure 7) . We then quantified HIV-1 DNA accumulation over time in DDX42-silenced and control cells. DDX42 depletion increased by 2-to 5-fold the accumulation of early and late reverse transcript products ( Figure 3A , B and C), as well as integrated provirus and 2-LTR circles at 48h post-infection ( Figure 3D and E). More than 85%

knockdown was achieved with both siRNAs targeting DDX42 ( Figure 3F ). These data suggested that DDX42 RNA helicase could inhibit the reverse transcription process and/or impact the stability of HIV-1 genome, leading to a decrease in viral DNA accumulation. We hypothesized that if that was the case, DDX42 should be found in close proximity to HIV-1 reverse transcription complexes during infection. In agreement with this, proximity ligation assay (PLA)

performed on MDMs infected with HIV-1 showed that DDX42 could indeed be found in close vicinity of Capsid ( Figure 3F and G).

We next examined the ability of DDX42 to inhibit infection by a range of primate lentiviruses including laboratory-adapted strains of HIV-1, HIV-1-transmitted founder strains, HIV-2 and simian immunodeficiency virus derived from the rhesus macaque (SIVMAC). TZM-bl cells were transfected with DDX42-targeting or scramble siRNAs and infected with VSV-G-pseudotyped lentiviruses. Infection efficiencies were monitored after 24h by measuring β-galactosidase activity ( Figure 3H ). DDX42 depletion increased infection levels with all the tested HIV-1 strains to the same extent than what was observed with HIV-1 NL4-3 (i.e. 3-to 5-fold). HIV-2rod10 and SIVMAC infection efficiencies were also slightly improved in the absence of DDX42 (by about 2-fold). The analysis was then extended to two non-primate lentiviruses, the equine infectious anaemia virus (EIAV) and feline immunodeficiency virus (FIV), using GFP-coding LVs derived from these viruses in comparison to HIV-1 and HIV-2 LVs (SI Figure 8 ). DDX42 antiviral activity appeared less potent on HIV-1 LVs compared to replication-competent, full-length HIV-1, which might suggest that viral components, absent in LVs, could be playing a role in DDX42-mediated HIV-1 inhibition.

Nevertheless, DDX42 depletion appeared to increase HIV-1, HIV-2 and FIV LV infection to the same extent, i.e. by about 2-fold, whereas EIAV infection was less impacted by DDX42 (SI Figure   8 ). We extended this study to the gammaretrovirus murine leukaemia virus (MLV) and observed that DDX42 depletion led to an increase in infection with GFP-coding MLV vectors ( Figure 3I ).

These results strongly support a general antiviral activity of DDX42 against retroviruses.

DDX42 can be found in the cytoplasm but is predominantly located in the nucleus (SI Figure 9 ; Uhlmann-Schiffler et al., 2009; Zyner et al., 2019) . Considering that DDX42 showed a broad activity against retroviruses and seemed to act at the level of reverse transcription, we sought to investigate whether DDX42 could inhibit retrotransposons. Long Interspersed Nuclear Elements (LINE)-1 are non-LTR retrotransposons, which have been found to be active in the germ line (Branciforte and Martin, 1994; Ergün et al., 2004; Trelogan and Martin, 1995) and in some somatic cells (Belancio et al., 2010; Muotri et al., 2005; Rangwala et al., 2009) . Interestingly, DDX42 was identified among the suppressors of LINE-1 retrotransposition through a genome-wide screen in K562 cells, although not further characterized (Liu et al., 2018) . To confirm that DDX42 could inhibit LINE-1 retrotransposition, HEK293T cells were co-transfected with two different, GFPexpressing LINE-1 plasmids (RPS or LRE3) or an inactive LINE-1 (JM111) together with a DDX42-or a control (Firefly)-expressing plasmid ( Figure 3J ). GFP-LINE-1 retrotransposition was quantified by flow-cytometry 7 days post-transfection (Moran et al., 1996) . Because the GFP cassette is cloned in antisense and disrupted by an intron in this reporter system, GFP is only expressed after LINE-1 transcription, splicing and Orf2p-mediated reverse-transcription and integration into the host genome (Moran et al., 1996) . Considering that most LINE-1 replication cycles lead to truncations and defective integrations (Gilbert et al., 2005) , GFP expression derived from a new integration is a relatively rare event and, as expected, the percentage of GFP+ cells observed was very low ( Figure 3J ) (Figure 4) . Strikingly, DDX42 depletion did not have an impact on IAV or VSV replication ( Figure 4A and B), thereby confirming that manipulating DDX42 expression did not have a broad and unspecific impact on target cells. However, depletion of endogenous DDX42

increased infection with ZIKV, CHIKV and SARS-CoV-2, and had a particularly high impact on the latter two (up to 1 log and 3 log increase in infection efficiency in DDX42-depleted cells in comparison to control cells, for CHIKV and SARS-CoV-2, respectively, Figure 4D and F). Of note, silencing efficiency was similar in the two types of target cells used here ( Figure 4E and G).

Interestingly, DDX42 was recently identified as a potential inhibitor of SARS-CoV-2 replication in a whole-genome CRISPR/Cas9 screen in simian Vero E6 cells (Wei et al., 2020) , supporting our observations that endogenous DDX42 potently inhibits the replication of this highly pathogenic coronavirus. Taken together, our data showed that DDX42 is a broad inhibitor of viral infections, albeit presenting some specificity. Further work is now warranted to explore in depth the breadth of DDX42 antiviral activity. . In contrast, our study revealed broad activity of endogenous DDX42 among retroviruses and retroelements, which was observed in various cell types, including primary CD4+ T cells. Interestingly, our PLA assays showed a close proximity between DDX42 and HIV-1 Capsid, which is a viral protein recently shown to remain associated with reverse transcription complexes until proviral DNA integration in the nucleus (Burdick et al., 2020; Dharan et al., 2020; Peng et al., 2014) . This observation could suggest a direct mode of action on viral ribonucleoprotein (RNP) complexes. Interestingly, we observed that DDX42 was able to inhibit viruses from other families, which possess different replication strategies, including SARS-CoV-2 and CHIKV. However, DDX42 did not have an impact on all the viruses we tested, reminiscent of broad-spectrum antiviral inhibitors such as MxA, which show some specificity (Haller et al., 2015) . DDX42 is known to be a non-processive helicase, which also possesses RNA annealing activities and the ability to displace RNA-binding proteins from single-stranded RNAs (Uhlmann-Schiffler et al., 2006) . Moreover, DDX42 binds G-quadruplexes (Zyner et al., 2019) , which are secondary structures found in cellular and viral nucleic acids and involved in various processes, such as transcription, translation and replication (Fay et al., 2017; Ruggiero and Richter, 2018) . All these known activities of DDX42 would be consistent with a potential role in RNP remodeling (Uhlmann-Schiffler et al., 2006; Will et al., 2002) . Nonetheless, further investigation will be needed to determine whether DDX42 acts directly by altering viral RNPs, and, if that's the case, what are the determinants for viral RNP recognition. In conclusion, this work highlights the importance of understanding the mechanism of action of DDX42 RNA helicase and its contribution to the control of RNA virus replication, an understanding which may contribute to the development of future antiviral interventional strategies.

plasmids were a gift from Prof. F. Zhang (Addgene #52962, #52963, and #1000000048, respectively ). LVs coding for sgRNAs targeting the candidate genes and bearing HIV-1 NL4-3, IIIB and HIV-2 proviral clones have been described (Adachi et al., 1986; Simon et al., 1995; Schaller et al., 2011) , as well as the transmitted founder HIV-1 molecular clones CH077.t, CH106.c, REJO.c (gifts from Prof. B. Hahn, (Ochsenbauer et al., 2012) ) and HIV-2ROD10 and SIVMAC239 (Ryan-Graham and Peden, 1995; Gaddis et al., 2004) . pBlaM-Vpr and pAdVAntage have been described (Cavrois et al., 2002) . GFP-coding HIV-1 based LV system (i.e. p8.91 HIV-1 Gag-Pol, pMD.G, and GFP-coding minigenome), and HIV-2, FIV, and EIAVderived, GFP coding LVs, as well as MLV-derived, GFP coding retroviral vectors have all been described (Naldini et al., 1996; Bainbridge et al., 2001) , (O'Rourke et al., 2002; Saenz et al., 2005) . The LINE-1 plasmid 99 RPS-GFP PUR (pRPS-GFP), 99 RPS-GFP JM111 PUR (pJM111) and pLRE3-GFP were developed by Prof. Kazazian's lab (Moran et al., 1996; Ostertag et al., 2000; Goodier et al., 2012) . was added at 1000 U/mL for 16-24h prior to virus infection or RNA extraction, and AZT and 3TC

(AIDS reagent program) at 10 µM for 2 h prior to infection.

Sang, under agreement n°21PLER2019-0106. Peripheral blood mononuclear cells (PBMCs) were isolated by centrifugation through a Ficoll® Paque Plus cushion (Sigma-Aldrich). Primary human CD4+ T cells and monocytes were purified by positive selection using CD3 and CD14

MicroBeads, respectively (Miltenyi Biotec), as previously described (Goujon et al., 2013a) . HIV-1 Renilla and NL4-3 HIV-1 were produced by standard PEI transfection of HEK293T. When indicated, pMD.G was cotransfected with the provirus at a 3:1 ratio. The culture medium was changed 6h later, and virus-containing supernatants were harvested 42h later. Viral particles were filtered, purified by ultracentrifugation through a sucrose cushion (25% weight/volume in Tris-NaCl-EDTA buffer) for 75 min at 4°C and 28,000 rpm using a SW 32 TI rotor (Beckman Coulter), resuspended in serum-free RPMI 1640 or DMEM medium and stored in small aliquots at -80°C.

β-lactamase-Vpr (BlaM-Vpr)-carrying viruses, bearing the wild-type Env, were produced by cotransfection of HEK293T cells with the NL4-3/Nef-IRES-Renilla provirus expression vector, pBlaM-Vpr and pAdVAntage at a ratio of 4:1:0.5, as previously described (Cavrois et al., 2002) . Viral particles were titrated using an HIV-1 p24 Gag Alpha-Lisa kit and an Envision plate reader (Perkin Elmer) and/or by determining their infection titers on target cells.

wild-type and/or VSV-G pseudotyped-HIV-1, target cells were plated at 2.5 x 10 4 cells per well in 96-well plates or at 2 x 10 5 cells per well in 12-well plates and infected for 24-48 h before lysis and Renilla (and Firefly) luciferase activity measure (Dual-Luciferase® Reporter Assay System Promega) or fixation with 2% paraformaldehyde (PFA)-PBS, permeabilization (Perm/Wash buffer, BDBiosciences) and intracellular staining with the anti-p24 Gag KC57-FITC antibody (Beckman Coulter), as described previously (Goujon and Malim, 2010) . For TZM-bl assays, the bgalactosidase activity was measured using the Galacto-Star™ system (ThermoFisher Scientific). (Corman et al., 2020) .

BlaM-Vpr assay for HIV-1 entry. These assays were performed as described previously (Goujon and Malim, 2010) . Briefly, 2 pNL4-3 or pTOPO-2LTR (generated by pTOPO cloning of a 2-LTR circle junction amplified from NL4-3 infected cells, using oHC64 and U3-reverse primers into pCR™2.1-TOPO™) were diluted in 20 ng/ml of salmon sperm DNA to create dilution standards used to quantify relative cDNA copy numbers and confirm the linearity of all assays.

Proximity Ligation assay. The proximity ligation assays were performed using the Duolink® in situ Detection Reagents (Sigma-Aldrich, DUO92014). For this, MDMs were plated in 24-well plates with coverslips pre-treated with poly-L-lysin (Sigma-Aldrich) and infected with 1 µg p24 Gag of HIV-1 NL4-3 (Ba-L Env) or mock-infected. 24 h later, the cells were fixed with 4% paraformaldehyde in PBS1X for 10 min, washed in PBS1X and permeabilized with 0.2% Triton X-100 for 10 min. After a couple of washes in PBS1X, either NGB buffer (50 mM NH4Cl, 2% goat serum and 2% bovine serum albumin in PBS) or Duolink® blocking solution was added for 1h.

Cells were incubated with AG3.0 mouse anti-Capsid antibody obtained from the National Institutes of Health (NIH) AIDS Reagent Program (#4121) and anti-DDX42 rabbit antibody (HPA023571, Sigma-Aldrich) diluted in NGB buffer or in Duolink® blocking solution for 1h. After 2 washes in PBS1X, the cells were incubated with the DUOLINK® in situ PLA® Probe Anti-rabbit minus (DUO92006) and DUOLINK® in situ PLA® Probe Anti-mouse plus (DUO92001) for 1h at 37°C. After 2 washes in PBS1X, the ligation mix was added for 30 min at 37°C. After 2 washes in PBS1X, the cells were incubated with the amplification mix for 100 min at 37°C. Finally, the cells were washed twice with PBS1X and stained with Hoechst at 1 µg/mL for 5 min, washed again and the coverslips mounted on slides in Prolong mounting media (ThermoFisher Scientific). Zstack images were acquired using an LSM 880 confocal microscope (ZEISS) using a 63x lens.

PLA punctae quantification was performed using the FIJI software (Schindelin et al., 2012) .

Briefly, maximum z-projections were performed on each z-stack and the number of nuclei per field were quantified. Then, by using a median filter and thresholding, PLA punctae were isolated and quantified automatically using the Analyse Particles function. To obtain a mean number of dots per cell, the number of PLA dots per field were averaged by the number of nuclei. For representative images, single cells were imaged using a LSM880 confocal microscope coupled with an Airyscan module. Processing of the raw Airyscan images was performed on the ZEN Black software.

Immunoblot analysis. Cell pellets were lysed in sample buffer (200 mM Tris-HCl, pH 6.8, 5.2% SDS, 20% glycerol, 0.1% bromophenol blue, 5% β-mercaptoethanol), resolved by SDS-PAGE and analysed by immunoblotting using primary antibodies specific for human DDX42 (HPA023571, Sigma-Aldrich) and Actin (mouse monoclonal A1978, Sigma-Aldrich), followed by secondary horseradish peroxidase-conjugated anti-mouse or anti-rabbit immunoglobulin antibodies and chemiluminescence (Bio-Rad). Images were acquired on a ChemiDoc™ gel imaging system (Bio-Rad).

We have described previously IAV NanoLuciferase reporter virus generation (Doyle et al., 2018) .

Stocks were titrated by plaque assays on MDCK cells. IAV challenges were performed in serum- ZIKV production and infection. The nanoluciferase expressing ZIKV construct has been described (Mutso et al., 2017) . The corresponding linearized plasmid was transcribed in vitro using the SP6 mMESSAGE mMACHINE™ (Thermofischer Scientific) and HEK293T cells were transfected with the transcribed RNA. After 7 days, supernatants were harvested, filtered and stock titers were determined by plaque assays on Vero cells. For infections, 2.5 x 10 4 cells per well in 96-well plates were infected, at the indicated MOIs. 24h after infection, cells were lysed and Nanoluciferase activity was measured using the Kit Nano Glo luciferase Assay (Promega).

CHIKV production and infection. The Gaussi luciferase coding CHIKV construct has been described (Pohjala et al., 2011) . The linearized plasmid coding CHIKV genome was transcribed with the T7 mMESSAGE mMACHINE kit (Thermofischer Scientific) and 5 x 10 5 HEK293T were transfected with 1-4 µg of transcribed RNA, using Lipofectamine 2000 (Thermofischer Scientific).

After 24h, supernatants were harvested, filtered and viruses were then amplified on baby hamster kidney (BHK21) cells. Stock titers were determined by plaque assays on Vero cells. For infections, 2.5 x 10 4 cells per well in 96-well plates were infected at the indicated MOIs. 24h after infection, cells were lysed and Gaussia luciferase activity was measured using the Pierce™ Gaussia Luciferase Flash Assay Kit (Thermofischer Scientific).

The SARS-CoV-2 BetaCoV/France/IDF0372/2020 isolate was supplied by Pr. Sylvie van der Werf and the National Reference Centre for Respiratory Viruses hosted by Institut Pasteur (Paris, France). The virus was amplified in Vero E6 cells (MOI 0,005) in serum-free media supplemented with 0,1 µg/ml L-1-p-Tosylamino-2-phenylethyl chloromethylketone (TPCK)-treated trypsin (Sigma-Aldrich). The supernatant was harvested at 72 h post infection when cytopathic effects were observed (with around 50% cell death), cell debris were removed by centrifugation, and aliquots stored at -80C. Viral supernatants were titrated by plaque assays in Vero E6 cells. Typical titers were 5.10 6 plaque forming units (PFU)/ml.

Infections of A549-ACE2 cells were performed at the indicated multiplicity of infection (MOI; as calculated from titers obtained in Vero E6 cells) in serum-free DMEM and 5% serum-containing DMEM, respectively. The viral input was left for the duration of the experiment and cells lysed at 48 h post-infection for RT-qPCR analysis.

The datasets generated during and/or analysed during the current study are available from the corresponding authors on reasonable request.

Requests for material should be addressed to Caroline Goujon or Olivier Moncorgé at the corresponding address above, or to Addgene for the plasmids with an Addgene number.

B.B. and CG designed the study, analysed the data and wrote the manuscript. B.B. and C.G. and MaGECK analyses, respectively. All authors have read and approved the manuscript.

The authors have no conflicts of interest to declare in relation to this manuscript. 

Supplementary

Production of acquired immunodeficiency syndrome-associated retrovirus in human and nonhuman cells transfected with an infectious molecular clone

Alpha interferoninduced antiretroviral activities: restriction of viral nucleic acid synthesis and progeny virion production in human immunodeficiency virus type 1-infected monocytes

In vivo gene transfer to the mouse eye using an HIV-based lentiviral vector; efficient long-term transduction of corneal endothelium and retinal pigment epithelium

Inhibition of human immunodeficiency virus (HIV) replication by HIV-trans-activated alpha 2-interferon

Somatic expression of LINE-1 elements in human tissues

Developmental and cell type specificity of LINE-1 expression in mouse testis: implications for transposition

Complex interplay between HIV-1 Capsid and MX2-independent IFNα-induced antiviral factors

HIV-1 uncoats in the nucleus near sites of integration

A sensitive and specific enzyme-based assay detecting HIV-1 virion fusion in primary T lymphocytes

Specific inhibition of viral protein synthesis in HIV-infected cells in response to interferon treatment

Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR

Nuclear pore blockade reveals that HIV-1 completes reverse transcription and uncoating in the nucleus

Am I ready for CRISPR? A user's guide to genetic screens

HIV-1 and interferons: who's interfering with whom?

The interferon-inducible isoform of NCOA7 inhibits endosome-mediated viral entry

Cell Type-specific Expression of LINE

Open Reading Frames 1 and 2 in Fetal and Adult Human Tissues

RNA G-Quadruplexes in Biology: Principles and Molecular Mechanisms

Further Investigation of Simian Immunodeficiency Virus Vif Function in Human Cells

Novel host restriction factors implicated in HIV-1 replication

Multiple Fates of L1 Retrotransposition Intermediates in Cultured Human Cells

MOV10 RNA Helicase Is a Potent Inhibitor of Retrotransposition in Cells

Characterization of the alpha interferon-induced postentry block to HIV-1 infection in primary human macrophages and T cells

Characterization of Simian Immunodeficiency Virus SIVSM/Human Immunodeficiency Virus Type 2 Vpx Function in Human Myeloid Cells

Human MX2 is an interferon-induced post-entry inhibitor of HIV-1 infection

Evidence for IFNα-induced, SAMHD1-independent inhibitors of early HIV-1 infection

Comprehensive Mutational Analysis of Yeast DEXD/H Box RNA Helicases Required for Small Ribosomal Subunit Synthesis

HIV-1 blocks the signaling adaptor MAVS to evade antiviral host defense after sensing of abortive HIV-1 RNA by the host helicase DDX3

Mx GTPases: dynamin-like antiviral machines of innate immunity

Recombinant human interferon alfa-A suppresses HTLV-III replication in vitro

Immunoproteasome activation enables human TRIM5α restriction of HIV-1

MX2 is an interferon-induced inhibitor of HIV-1 infection

Ultrafast and memory-efficient alignment of short DNA sequences to the human genome

The Sequence Alignment/Map format and SAMtools

MAGeCK enables robust identification of essential genes from genome-scale CRISPR/Cas9 knockout screens

Selective silencing of euchromatic L1s revealed by genome-wide screens for L1 regulators

The highly polymorphic cyclophilin A-binding loop in HIV-1 capsid modulates viral resistance to MxB

Cutadapt removes adapter sequences from high-throughput sequencing reads

Investigation of Influenza Virus Polymerase Activity in Pig Cells

High frequency retrotransposition in cultured mammalian cells

Somatic mosaicism in neuronal precursor cells mediated by L1 retrotransposition

Reverse genetic system, genetically stable reporter viruses and packaged subgenomic replicon based on a Brazilian Zika virus isolate

In vivo gene delivery and stable transduction of nondividing cells by a lentiviral vector

Generation of transmitted/founder HIV-1 infectious molecular clones and characterization of their replication capacity in CD4 T lymphocytes and monocyte-derived macrophages

A virus-packageable CRISPR screen identifies host factors mediating interferon inhibition of HIV

Comparison of gene transfer efficiencies and gene expression levels achieved with equine infectious anemia virus-and human immunodeficiency virus type 1-derived lentivirus vectors

Determination of L1 retrotransposition kinetics in cultured cells

Quantitative microscopy of functional HIV post-entry complexes reveals association of replication with the viral capsid

Inhibitors of Alphavirus Entry and Replication Identified with a Stable Chikungunya Replicon Cell Line and Virus-Based Assays

Many LINE1 elements contribute to the transcriptome of human somatic cells

A Vesicular Stomatitis Virus Replicon-Based Bioassay for the Rapid and Sensitive Determination of Multi-Species Type I Interferon

Characterization of the ATPase and unwinding activities of the yeast DEADbox protein Has1p and the analysis of the roles of the conserved motifs

G-quadruplexes and G-quadruplex ligands: targets and tools in antiviral therapy

Both Virus and Host Components Are Important for the Manifestation of a Nef-Phenotype in HIV-1 and HIV-2

Restriction of feline immunodeficiency virus by Ref1, Lv1, and primate TRIM5alpha proteins

Improved vectors and genome-wide libraries for CRISPR screening

HIV-1 capsid-cyclophilin interactions determine nuclear import pathway, integration targeting and replication efficiency

Fiji: an open-source platform for biologicalimage analysis

Genome-scale CRISPR-Cas9 knockout screening in human cells

High-throughput functional genomics using CRISPR-Cas9

Complementation of vif-defective human immunodeficiency virus type 1 by primate, but not nonprimate, lentivirus vif genes

Controls Use of the HIV A4/5 Splice Acceptor Cluster and Is Essential for Efficient Replication of HIV

DDX5 potentiates HIV-1 transcription as a co-factor of Tat

The DEAD-box helicase DDX3 substitutes for the cap-binding protein eIF4E to promote compartmentalized translation initiation of the HIV-1 genomic RNA

Tightly regulated, developmentally specific expression of the first open reading frame from LINE-1 during mouse embryogenesis

Ddx42p--a human DEAD box protein with RNA chaperone activities

The DEAD box protein Ddx42p modulates the function of ASPP2, a stimulator of apoptosis

Genome-wide CRISPR screens reveal host factors critical for SARS-CoV-2 infection

Characterization of novel SF3b and 17S U2 snRNP proteins, including a human Prp5p homologue and an SF3b DEAD-box protein

Identification of RNA helicases in human immunodeficiency virus 1 (HIV-1) replication -a targeted small interfering RNA library screen using pseudotyped and WT HIV-1

FastQ Screen: A tool for multi-genome mapping and quality control

Role of MxB in Alpha Interferon-Mediated Inhibition of HIV

Requirement of DDX3 DEAD Box RNA Helicase for HIV-1 Rev-RRE Export Function

Genetic interactions of G-quadruplexes in humans

Indirect immunofluorescence analysis of endogenous DDX42 in primary MDMs. MDMs were fixed, endogenous DDX42 and the nuclei were visualized using DDX42-specific antibodies and Hoechst staining, respectively, and confocal microscopy

We wish to thank Tom Doyle and Chad Swanson for their useful comments on the manuscript,