Vesicular Stomatitis Virus Transcription Is Inhibited by TRIM69 in the Interferon-Induced Antiviral State

Interferons are important antiviral cytokines that work by inducing hundreds of host genes whose products inhibit the replication of many viruses. While the antiviral activity of interferon has long been known, the identities and mechanisms of action of most interferon-induced antiviral proteins remain to be discovered. We identified gene products that are important for the antiviral activity of interferon against vesicular stomatitis virus (VSV), a model virus that whose genome consists of a single RNA molecule with negative-sense polarity. We found that a particular antiviral protein, TRIM69, functions by a previously undescribed molecular mechanism. Specifically, TRIM69 interacts with and inhibits the function of a particular phosphoprotein (P) component of the viral transcription machinery, preventing the synthesis of viral messenger RNAs.

so-called "antiviral state," in which the replication of many viruses is blocked (2,3). Some ISGs have been shown to inhibit specific processes in viral replication or induce the destruction or depletion of viral proteins or RNA. However, the function of the majority of individual ISGs and precisely how they inhibit the replication of particular viruses remain unknown (4).
Vesicular stomatitis virus (VSV), a prototypic member of the order Mononegavirales and an important animal pathogen, is highly sensitive to inhibition by type I IFN. Like that of many RNA viruses, VSV replication occurs within specialized compartments in the cytoplasm of infected cells (5,6). Compartmentalization may help to shield viral components from detection by cytosolic sensors or antiviral proteins that might otherwise increase IFN production or directly interfere with viral replication. For rhabdoviruses, such as VSV and rabies virus (RabV), and other cytoplasmically replicating negative-strand RNA viruses, replication compartments are not circumscribed by a membrane (6)(7)(8)(9). Instead, replication components form inclusions that manifest features characteristic of phase-separated liquid compartments, such as P bodies and nucleoli (10,11). Three VSV proteins, namely, the nucleoprotein (N), which coats the viral RNA, the large protein (L), which possesses all the viral enzymes necessary for transcription, and the phosphoprotein (P), which binds to both N and L to stimulate RNA synthesis, are necessary and sufficient for the assembly of phase-separated replication compartments (10). While not required for compartment formation, negativestrand viral RNA is located, replicated, and transcribed within these compartments once infection is established (6). However, the initial "pioneer" round of transcription, in which viral mRNAs are transcribed from a single incoming negative-strand viral genome, employs N, P, and L proteins that enter the cell as components of the incoming viral particle and, thus, occurs prior to the formation of phase-separated replication compartments.
Among the known antiviral proteins, a number of tripartite motif (TRIM) proteins have been shown to interfere directly with key steps in the life cycles of widely divergent viruses or exert indirect inhibition as regulators of antiviral signaling (12). While a variety of very distinct mechanisms and functions have been ascribed to TRIM proteins in this context, a characteristic feature of TRIM proteins is their shared architecture. Specifically, TRIM proteins form antiparallel dimers, driven by a central coiled-coil domain, that constitute one defining feature of the tripartite motif (13)(14)(15). At least some TRIM proteins also form higher-order multimers, mediated by interactions between N-terminal RING and/or B-box domains (16,17), that are also defining features of the tripartite motif. Typically, SPRY or other protein domains situated at the TRIM protein C terminus enable interactions with viral or cellular targets (12). The propensity of TRIM proteins to form high-order multimeric structures thereby allows polyvalent interactions with targets.
Herein, we describe a loss-of-function screen to identify ISGs that are mediators of the anti-VSV activity of IFN-␣. We show that the products of multiple ISGs, including previously unidentified antiviral proteins, contribute to the overall activity of IFN-␣. Among these proteins, we identify a poorly characterized TRIM protein, TRIM69, as an inhibitor of VSV replication. We show that TRIM69 inhibits VSV replication through a previously unanticipated mechanism of action. Specifically, we find that higher-order TRIM69 multimers target a specific sequence in VSV P. In so doing, TRIM69 inhibits viral transcription and the formation of VSV replication compartments, resulting in profound reduction of viral RNA synthesis and inhibition of viral replication.
(This article was submitted to an online preprint archive [18].)

Identification of ISGs that mediate the anti-VSV activity of IFN-␣.
To identify genes responsible for the antiviral state, specifically those responsible for inducing VSV (Indiana serotype [VSV IND ]) resistance, we selected a subclone of HT1080 cells in which IFN potently inhibited the replication of a recombinant VSV IND engineered to carry a nanoluciferase (nLuc) reporter gene [VSV IND (nLuc)]. We also designed a small interfer-ing RNA (siRNA) library containing siRNA pools (Dharmacon SMARTpools containing 4 individual siRNA duplexes) representing the 400 most strongly upregulated genes among a panel of cell lines, along with 18 siRNA controls (Table 1). We transfected HT1080 cells in 96-well plates with the arrayed siRNA library and treated the cells with 10 U/ml IFN-␣. The following day, cells were infected at a low multiplicity of infection (MOI) (0.01) with VSV IND (nLuc). After a further day, luciferase activity was measured, to identify siRNA pools that were able to enhance spreading VSV IND replication in the presence of IFN-␣ (Fig. 1A).
Anti-VSV IND activity is associated with high-order TRIM69 multimerization. TRIM69 is a member of the large family of TRIM proteins, several of which have been shown to exhibit direct or indirect antiviral activity (12). Like other TRIM proteins, TRIM69 protein forms dimers that are held together by a coiled-coil domain, configured in antiparallel orientation (15). In addition, for at least some TRIM proteins, the antiparallel dimers are assembled into higher-order multimeric structures, which are held together via a second dimer or trimer interface, located in the RING or B-box domains, respectively (16,17). Among other TRIM proteins, TRIM25 contains a dimeric interface in its RING domain and also shares the highest level of sequence similarity with TRIM69 (Fig. 4A). We used the previously determined crystal structure of a TRIM25 RING domain dimer (17) to identify amino acids at the RING domain dimer interface (V95, L96, and L99 in TRIM69, analogous to V68, L69, and V72 in TRIM25) ( Fig. 4A and B) whose mutation might disrupt the TRIM69 RING dimer interface and, perhaps, higher-order TRIM69 multimerization (Fig. 4B).

TRIM69 Inhibits VSV Transcription
Journal of Virology notion that it assembles into high-order structures. Moreover, Western blot analysis of cell lysates generated after treatment of cells with the protein cross-linker EGS [ethylene glycol bis(succininic acid N-hydroxysuccinimide ester)] showed that TRIM69 formed dimers and, apparently, higher-order multimers (Fig. 4E). Notably, mutation of the predicted RING domain dimer interface (a V95A L96A double mutant [bearing changes of V to A at position 95 and L to A at position 96] and an L99A single mutant) resulted in TagRFP/TRIM69 proteins that formed cross-linkable dimers but not higher-order multimers (Fig. 4E) and exhibited a diffuse rather than filamentous distribution in the cell cytoplasm (Fig. 4D). Notably, V95A L96A and L99A mutations abolished the ability of TRIM69 to inhibit VSV IND (nLuc) replication, suggesting that higher-order multimerization is necessary for antiviral activity (Fig. 4F).
Recruitment of TRIM69 to sites of VSV replication and inhibition of replication compartment formation. Like many RNA viruses, VSV partitions its replication machinery into specialized compartments within which RNA synthesis occurs (6)(7)(8)(9). In the case of VSV, recent work has shown that these compartments are liquid inclusions that are not bounded by membranes but instead exhibit characteristics of phase separation (10). Replication compartments are conveniently labeled and localized using VSV clones modified to append a fluorescent protein to the amino terminus of P. We infected cells with VSV IND (NeonGreen/P) and visualized replication compartments using NeonGreen green fluorescent protein (GFP), along with fluorescence in situ hybridization (FISH) probes directed to the negative-strand viral RNA (Table 2). This analysis showed that the presence of TRIM69 profoundly attenuated the formation of replication compartments marked by the P protein and negative-strand RNA (Fig. 5). Additionally, 3-D-SIM imaging of cells expressing mScarlet/TRIM69 and infected with VSV IND (NeonGreen/P) revealed that the smaller P accumulations that were observed in TRIM69-expressing cells were typically colocalized with the mScarlet/TRIM69 filaments (Fig. 6). Indeed, many of the P accumulations appeared to adopt an elongated, almost filamentous structure, different in shape and size from the compartments typically observed in VSV-infected cells, and coincident with TRIM69 filaments (Fig. 6). Overall, these data suggested that a viral or cellular component governing the formation of viral replication compartments associates with TRIM69 and that this association ultimately inhibits replication compartment formation.
The VSV phosphoprotein (P) is the viral determinant of TRIM69 sensitivity. To determine how TRIM69 inhibits VSV replication, we selected mutant VSV derivatives that were resistant to TagRFP/TRIM69. We infected HT1080 cells expressing TagRFP/ TRIM69 with VSV IND (eGFP) or a VSV IND clone in which eGFP was appended to P [VSV IND (eGFP/P)] at a high multiplicity (MOI ϭ 10). Three growing plaques each were picked for VSV IND (eGFP) and VSV IND (eGFP/P) and amplified on HT1080 cells expressing TagRFP/TRIM69. These viruses infected HT1080 cells with equivalent efficiency whether TagRFP/TRIM69 was induced or not ( Fig. 7A and B). Sequencing of the viral genome of these TRIM69-resistant (TR) viruses showed that all six encoded a nonsynonymous mutation within a short peptide sequence (amino acids 66 to 71) within P [P(66 -71)] (Fig. 8A). For 2/6 viruses, the only mutations present were nonsynonymous changes in the P(66 -71) peptide; in another 2/6 viruses, additional synonymous mutations were present; while in a further 2/6 viruses, additional nonsynonymous mutations were found. Based on these findings, we concluded that the P(66 -71) mutations were likely responsible for the resistance to TRIM69 (Fig. 8A). This peptide sequence is within a region of P that contacts the globular connector, methyl transferase, and C-terminal domain (CTD) of L and is also responsible for stimulating polymerase activity (the L-stimulatory region [LSR]) ( Fig. 8A) (20,21). Although located in a region crucial for efficient viral RNA synthesis, the TR mutations did not affect replication of VSV IND (eGFP) or VSV IND (eGFP/P) in Vero cells (Fig. 8B). Notably, the corresponding amino-acid-66 to -71 sequence in the P protein of VSV NJ is different in 5/6 amino acid positions (Fig. 8A), providing a potential explanation for the intrinsic resistance of VSV NJ to TRIM69.
TRIM69 associates with VSV IND P. We next investigated how the TR mutations in P exerted their effects. While wild-type (WT) TagRFP/TRIM69 effectively prevented the formation of VSV IND (eGFP/P) replication compartments, the two VSV IND (eGFP/P) TR mutants, SVIND(eGFP/P)_TR1(D70Y) and VSVIND(eGFP/P)_TR3(E67G), bearing a D70Y and an E67G mutation, respectively, formed prominent replication compartments that showed no association with TRIM69 filaments or accumulations (Fig. 9). Strikingly, the multimerization-defective mutants of TRIM69 (V95A L96A and L99A mutations), which did not exhibit antiviral activity and were ordinarily diffusely distributed in the cytosol (Fig. 4D), were nearly completely relocalized to replication compartments in VSV IND (NeonGreen/P)-or VSV IND (eGFP/P)-infected cells (Fig. 10A). Thus, TRIM69 recruitment to replication compartments was not itself sufficient to inhibit VSV IND replication. Moreover, the dramatic redistribution of TRIM69(L99A) to replication compartments in VSV IND -infected cells was completely absent in VSV IND (eGFP/P)_TR1(D70Y)-and VSV IND (eGFP/P)_TR3(E67G)-infected cells (Fig. 10A). Together, these data indicate that specific sequences in P are necessary for the recruitment of TRIM69 to the VSV IND replication machinery and vice versa.  1  TCT CTT GAC TGT AAC AGA CA  GGC AAG TAT GCT AAG TCA GA  2  GGA ACT ACG ACT GTG TTG TC  GCA AGG CCT AAG AGA GAA GA  3  ATC CTC ATT TGC AGG AAG TT  GCG AAA AGA GCA GTC ATG TC  4  AAG TAA TCT GCC GGG TAT TC  GGA TTG ACG ACT AAT GCA CC  5  GAG GAA TCT CCT TTG ATT TT  ACA CTC CAG ATG ATA GTA CC  6  GGT AGA CAT ATC CTC TTA GA  CGA CTT GGC ACA ACA GTT TT  7  TAC ATT TCC GGA TTT GAG GC  TTG TAC GCT TAT GCA GTA GG  8  GTA GCT GTT GAC ATG TAT GA  ATC TCT TAC TAC AGC AGG TT  9 ATG  TAC TTG TAT GG  36  AAA CTG TTG TGC CAA GTC GG  GCC TCA AAT CCG GAA ATG TA  37  CGG TAC TAT CAT CTG GAG TG  CTA AGA GGA TAT GTC TAC CA  38  CGG TGC ATT AGT CGT CAA TC  GAG ATT CCT CTT TAC ATC AA  39  GTG ACA TGA CTG CTC TTT TC  AGT GGA ATA CCC GGC AGA TT  40  ATT GTC TTC TCT CTT AGG CC  AAC TTC CTG CAA ATG AGG AT  41  TTC TGA CTT AGC ATA CTT GC  GAC AAC ACA GTC GTA GTT CC  42 TGT CTG TTA CAG TCA AGA GA TRIM69 Inhibits VSV Transcription Journal of Virology To determine whether P was sufficient for TRIM69 recruitment, we expressed eGFP/P by transfection, in the absence of any other viral proteins. In contrast to the situation in VSV IND -infected cells, eGFP/P expressed alone was diffusely distributed throughout the cytoplasm (Fig. 10B). However, in TagRFP/TRIM69-expressing cells, eGFP/P was recruited to the TagRFP/TRIM69 accumulations and the two proteins colocalized extensively ( Fig. 10B and C). Again, this colocalization was dependent on the viral determinant of TRIM69 sensitivity, as there was no colocalization between TagRFP/ TRIM69 and eGFP/P(E67G) (Fig. 10B and C).
To determine whether P physically associated with TRIM69, we generated glutathione S-transferase (GST) proteins fused to a region of P encompassing the LSR domain (amino acids 32 to 107). We also constructed GST/P fusion protein-containing mutant LSR domains from VSV IND (eGFP)_TR1(E69K) and from VSV IND (eGFP/P)_TR3(E67G) (Fig.  8A). The GST/P fusion proteins were coexpressed with Cherry/TRIM69 in 293T cells and then purified from cell lysates with glutathione-Sepharose. Cherry/TRIM69 was nearly undetectable in clarified cell lysates, presumably due to its propensity to form higherorder multimers that were poorly soluble in nondenaturing detergents (Fig. 10D). Nevertheless, GST/P coprecipitated Cherry/TRIM69 such that it was highly enriched in precipitated fractions (Fig. 10D). Conversely, GST/P proteins bearing the TR mutations (E69K or E67G) did not precipitate detectable amounts of Cherry/TRIM69 (Fig. 10D). Thus, the isolated LSR domain was sufficient for the physical association of P with TRIM69.
Mechanism of VSV replication inhibition by TRIM69. The aforementioned data provide evidence that TRIM69 physically associates with P and ultimately inhibits VSV IND replication compartment formation. However, the assembly of replication compartments requires multiple prior steps, each of which is a potential target of TRIM69. For example, because TRIM proteins can act as ubiquitin ligases (12), it was conceivable that TRIM69 might mediate the destruction of one or more viral proteins. Alternatively, interaction with P could block replication through a nondegradative mechanism by inhibiting (i) an initial pioneer round of transcription that employs the incoming virion RNA genome as a template to generate viral mRNAs, (ii) the translation of these new viral mRNAs to generate N, P, and L proteins, or (iii) the assembly of newly synthesized N, P, and L proteins with full-length negative-strand RNAs. To determine whether TRIM69 targeted incoming viral proteins for degradation, we first infected cells with 35 S-labeled virions at a high MOI and monitored the levels of virion proteins in the presence of cycloheximide (CHX) to prevent new protein synthesis and VSV replication. No substantial difference in the decay of incoming viral proteins in the presence or absence of TRIM69 was detected (Fig. 11A). Next, we monitored the levels of P protein following transfection into cells in which TRIM69 expression was or was not induced. Despite obvious recruitment of eGFP/P to sites of TRIM69 concen-  (Fig. 10B), the overall levels of WT and TR mutant eGFP/P were equivalent and unaffected by TRIM69 (Fig. 11B). Finally, we found that WT and inactive (L99A) mutants of TRIM69 exhibited approximately equivalent levels of autoubiquitination (Fig. 11C). Overall, we found no evidence that TRIM69 drives VSV protein degradation, and the ubiquitin ligase activity of TRIM69 was unable to account for its antiviral activity. Next, we quantified new viral mRNA (pioneer) transcription from incoming virion RNA templates in cells treated with CHX to prevent protein synthesis and replication. A single-molecule FISH assay and a pool of oligonucleotide probes directed at the plus-strand N mRNA (Table 2) demonstrated a clear reduction in pioneer N mRNA transcription in cells expressing TRIM69 (Fig. 12A and B). In an alternative approach to measure VSV transcription, we labeled target cells in the presence of actinomycin D (to block host mRNA synthesis) with [ 32 P]orthophosphate. At 5 h after VSV IND (eGFP/P) infection, transcripts corresponding to L, G, and M mRNAs were clearly detectable (Fig.  12C). Transcripts corresponding to N and/or eGFP/P mRNAs were also detected but could not be distinguished from each other due to comigration (Fig. 12C). As expected, in the absence of CHX, mRNA synthesis and genome replication of VSV IND (eGFP/P) were inhibited by TRIM69 but not TRIM69(L99A), while VSV IND (eGFP/P)_TR3(E67K) mRNA synthesis and genome replication were insensitive to TRIM69. Importantly, when CHX was used to prevent protein synthesis and RNA replication, the presence of TRIM69 reduced the levels of all nascent VSV IND mRNA transcripts (Fig. 12C). The magnitude of the effect on pioneer transcript levels appeared greatest for the L mRNA, least for the N and P mRNAs, and of intermediate magnitude for M and G mRNAs (12C and D). The multimerization-defective TRIM69(L99A) mutant did not show an effect on pioneer mRNA synthesis (Fig. 12C). Thus, TRIM69 inhibited primary transcription of the incoming VSV IND virion RNA, with apparently greater effect on genes encoded near the 5= end of the negative-strand genome.

DISCUSSION
A number of ISGs, including Mx1, PKR, IFITM3, and tetherin, have previously been reported to inhibit VSV replication (22)(23)(24)(25). However, it was not known whether this list represents a complete catalogue of ISG proteins with anti-VSV activity. We found that a number of antiviral ISGs contribute to the induced antiviral state that prevents VSV IND replication in IFN-␣-treated cells. Among these, we found that TRIM69 has a previously undescribed mechanism of action, inhibiting VSV IND transcription by targeting the polymerase cofactor, P.
TRIM69 joins a growing list of TRIM proteins that have been shown to exhibit antiviral activity through various mechanisms (12). During the course of this work, TRIM69 itself was reported to inhibit dengue virus type 2 (DENV2) replication, albeit via a different mechanism from that described herein, namely, ubiquitin-induced degradation of the viral NS3 RNA helicase (26). While the manuscript was in preparation, another group also found that TRIM69 inhibits VSV IND replication (27) but were unable to reproduce the reported finding of TRIM69 antiviral activity against DENV2 (26). These authors also demonstrated that P is a crucial determinant of TRIM69 sensitivity and  components (28,29). Another TRIM protein, TRIM25, has been reported to promote ubiquitination of the RNA sensor RIG-I, thereby inducing binding to mitochondrial antiviral signaling (MAVS) protein and stimulation of IFN production (30), although this model has recently been challenged (31). TRIM25 is also an important cofactor of the zinc finger antiviral protein (ZAP), which senses and depletes CG-rich viral RNAs, although the mechanism that enables ZAP activity remains unclear (32). A variety of other TRIM proteins have been reported to inhibit viral replication directly or indirectly through less well characterized mechanisms (12). In the two aforementioned examples, high-order multimerization is crucial for activity. TRIM5 higher-order multimerization, driven by a B-box domain, facilitates the formation of a hexagonal lattice on the surface of incoming retroviral capsid, enabling polyvalent interaction between the capsid hexagonal lattice and a complementary hexagonal TRIM5 lattice (16,33). In this case, higher-order multimerization results in a more avid interaction between TRIM5 and its viral capsid protein target. For TRIM25, RING domain dimerization enables engagement of ubiquitin-conjugated E2 enzymes and higher-order assembly of the RIG-I signalosome (17). Herein, we found that the RING domain dimer interface, analogous to that found in TRIM25, was required for higher-order TRIM69 multimerization, the formation of TRIM69 filaments, and antiviral function. However, abolition of high-order TRIM69 multimerization by mutation of the RING domain dimer interface did not prevent recruitment into VSV IND replication compartments. Rather, recruitment of dimeric TRIM69 to replication compartments remained efficient but was inconsequential to VSV replication. Thus, for TRIM69, RING domain-mediated multimerization appeared to be required for antiviral activity but not target recognition. As RING domain dimerization might lead to E2 recruitment, as well as high-order multimer formation, it is not clear whether higher-order multimerization per se or downstream E2 recruitment is essential for TRIM69 activity. However, the lack

TRIM69 Inhibits VSV Transcription
Journal of Virology of effect of TRIM69 on incoming virion protein stability or on coexpressed P levels, coupled with the finding that the L99A mutant maintained ubiquitin ligase activity, argues that destruction of virion proteins is not central to the mechanism of action of TRIM69. Unfortunately, we were not able to identify a TRIM69 mutant that maintained higher-order multimer formation but abolished ubiquitination activity. We did not formally demonstrate that TRIM69 directly binds to P, and it is possible that P interacts with some bridging host protein(s) that is(are) bound by TRIM69. However, a direct interaction between TRIM69 and the LSR domain of P is the most likely molecular event underlying recognition and disruption of the viral transcription/ replication machinery. P is required for the interaction between L and the N-coated RNA template (20,34) and, thus, for initial transcription following viral entry, as well as for the formation of replication compartments (10). Given that P plays a pivotal, multifactorial role in VSV RNA synthesis and has no cellular homologs, it represents an attractive target for intrinsic immune defenses. Moreover, the pioneer round of transcription may represent a point of vulnerability when the number of viral targets is low and ISG-mediated inhibition might exert maximal effects. Nevertheless, subsequent rounds of transcription are thought to proceed by essentially the same mechanism as pioneer transcription, and it is possible that TRIM69 inhibits transcription at all stages of VSV IND RNA accumulation within an infected cell. With the caveat that overexpressed TRIM69 was used, we noted the formation of elongated filamentous accumulations of P, coincident with TRIM69 filaments in TRIM69-expressing cells, rather than the spherical droplets that normally characterize the phase-separated VSV replication compartments. This suggests that TRIM69 might inhibit replication compartment formation, in addition to its effects on transcription. Because the L99A TRIM69 mutant retained the ability to be recruited by P and localizes to the phase-separated replication compartments and yet did not inhibit VSV IND transcription, it appears that the interaction of TRIM69 with the LSR of P does not prevent functional P-L complex formation. While further study will be required to elucidate the molecular details of how TRIM69 recognizes and disrupts the VSV replication machinery, these findings reveal a new facet of the diverse ways in which IFNs control the replication of viruses.
Plasmid construction. The pLKO-derived doxycycline-inducible lentiviral expression vector was used as previously described (3). pLKO TagRFP or mScarlet N-terminally tagged human TRIM69 constructs comprising various species and mutants were cloned by overlapping PCR using SfiI restriction sides. pLKO myc-TRIM69 or mutant constructs were cloned using a forward primer containing a myc tag and SfiI restriction sites. Plasmids expressing WT or mutant TRIM69-3ϫmyc were generated in the pCR3.1 expression vector containing an EcoRI-XhoI-NotI multiple cloning site, followed by the addition of an in-frame 3ϫmyc tag using EcoRI and XhoI restriction sites. Plasmids expressing WT or mutant Cherry-TRIM69 were generated in the pCR3.1 Cherry EcoRI-XhoI-NotI background using EcoRI and XhoI restriction sites.
Sequences encoding VSV-P WT and E67G and E69K mutants (amino acids 32 to 107) were amplified from pCAGGS-eGFP/P plasmids and inserted in frame with the glutathione-S-transferase (GST) gene in the pCAGGS-GST expression plasmid using the following restriction enzymes and oligonucleotide primers: 5= EcoRI and GAGGAGGAATTCGCTGAAAAGTCCAATTATGAGTTG and 3= XhoI and CTCCTCCTCG AGCTAGTCCGAAGTAAATACAACATCCAC.
The plasmids used to produce Sendai virus were kindly provided by Benhur Lee. The recombinant Sendai virus (rSeV) clone provided contained eGFP and mutations in the F and M genes to allow trypsin-independent growth. A hammerhead ribozyme (Hh-Rbz) sequence was present between the T7 promoter and the start of the viral antigenome to enhance the rescue efficiency (39). The GFP gene that had been positioned between the N and P genes via duplication of the N-P intergenic region was replaced with nLuc. Briefly, the N and P regions were amplified with forward primer Pre-SbfI-for (5=-TGACCATGATTACGCCAAGCTTAA-3=) and reverse primer nLuc_SeV_rev (5=-GAAATCTTCGAGTGTGAA GACCATGCGGTAAGTGTAGCCGAAGCCGTG-3=) and forward primer nLuc-SeV_for (5=-CTGTGCGAACGCAT TCTGGCGTAATGAGATAGGAGGAATCTAGGATCA-3=) and reverse primer Post-SmaI-6860-rev (5=-GATGGTAG ATTGGGTCTCTCTGTG-3=), respectively. The nLuc gene (Promega) was amplified with forward primer SeV-nLuc_for (5=-CACGGCTTCGGCTACACTTACCGCATGGTCTTCACACTCGAAGATTTC-3=) and reverse primer SeV-nLuc_rev (5=-TGATCCTAGATTCCTCCTATCTCATTACGCCAGAATGCGTTCGCACAG-3=). After overlap extension PCR using the outermost primers, the fragments were inserted into the Sbf1-and SmaI-digested rSeV_GFP construct using Gibson assembly.
pVSV NJ (ϩ)-eGFP was constructed as described previously for plasmid pVSV1(ϩ), encoding VSV IND genomic RNA (40). Briefly, pVSV NJ (ϩ)-eGFP was assembled from plasmid made by reverse transcription-PCR of each of the VSV NJ genes of the Ogden strain and intergenic junctions by standard cloning techniques. These clones were assembled into a full-length cDNA and inserted between the bacteriophage T7 promoter and a cDNA copy of the self-cleaving ribozyme from the antigenomic strand of hepatitis D virus (HDV). The eGFP gene was inserted in the first position (in an additional transcription unit before N) as described for VSV IND (41).
Viruses. Plasmids encoding the full-length VSV IND genome (pVSV-FL), as well as individual VSV IND N, P, L, and G genes, were purchased from Kerafast [VSV-FLϩ(2) VSV plasmid expression vector system, catalog number EH1002] or were generated as described previously (40). VSV IND viruses were generated by infecting 293T cells with T7-expressing vaccinia virus (vTF7-3) at an MOI of 5, followed by transfection with pVSV plasmids and plasmids encoding VSV N, P, L, and G under the control of a T7 promoter. Supernatants were harvested 48 h posttransfection (h.p.t.), filtered (0.2 m) to remove the bulk of the vaccinia virus, and plaque purified on 293T cells. Plaque-purified virus was expanded on 293T cells, and cell culture supernatant was harvested, passed through a 0.2-m filter, and frozen in aliquots. Virus titers (PFU/ml) were determined by plaque formation using HT1080 or BHK21 cells. VSV encoding nanoluciferase (nLuc) was generated by inserting the nLuc-encoding sequences (from pNL1.1; Promega) into the pVSV plasmid between the envelope and L genes, along with appropriate VSV regulatory sequences. VSV derivatives encoding mNeonGreen were generated by fusing the mNeonGreen-encoding sequence to the N terminus of P. Rescue of replication-competent Sendai virus from transfected plasmids was done as previously described (39) with transfection into 293T cells using Lipofectamine LTX (catalog number 15338100; Invitrogen) according to the manufacturer's recommendations. Virus titers (PFU/ml) were determined by plaque formation using HT1080 target cells.
siRNA experiments. Amounts of 3 ϫ 10 3 HT1080 cells were plated in a 96-well plate, transfected with siRNA SMARTpools or the most efficient individual siRNA (Dharmacon) ( Table 1), treated with increasing concentrations of IFN-␣ or a fixed dose of 10 U/ml of IFN-␣ (catalog number 11200-2; PBL Assay Science) at 8 h.p.t., and harvested as described above.
Inducible expression of TRIM69. Inducible HT1080 cells were generated by transduction with an pLKO-derived vector as described previously (3), followed by selection with 1.25 g/ml puromycin (catalog number P8833-100MG; Sigma-Aldrich). Expression was induced in pLKO-transduced cell lines through an overnight treatment with 0.5 g/ml doxycycline hyclate (catalog number 324385; Sigma-Aldrich) prior to viral challenge.
smFISH. Single-molecule fluorescent in situ hybridization (smFISH) probes against both the plus and minus strands of VSV N were designed using the Stellaris Probe Designer, version 2.0 (Biosearch Technologies) ( Table 2). For each RNA strand, 41 (plus strand) or 42 (minus strand) oligonucleotide probes were synthesized by IDT to contain a 5= amino modifier (C6). The 5= amino-modified probes for each RNA were resuspended to 1.25 g/ml, pooled, and purified by three chloroform extractions followed by ethanol precipitation. Then, 50-g amounts of the pooled probes were labeled with ester-modified Alexa Fluor 488 or Alexa Fluor 549 using the Alexa Fluor 488 oligonucleotide amine labeling kit (catalog number A20191; Thermo Fisher). After labeling, the pooled probes were ethanol precipitated, resuspended in RNase-free water, and purified via the Oligo Clean & Concentrator kit from Zymo Research (catalog number D4060). The pooled probes were eluted in RNase-free Tris-EDTA (TE), pH 8.0 (catalog number AM9849; Ambion), and adjusted to a final concentration of 12.5 M. For FISH, 3 ϫ 10 4 HT0180-myc/TRIM69 cells were seeded onto gelatin-coated, 8-chambered, no. 1.5 borosilicate glass-bottom slides (catalog number 155409; LabTek). Doxycycline-treated or untreated cells were pretreated for 30 min with 100 g/ml of cycloheximide (catalog number C4859; Sigma-Aldrich) and infected at an approximate MOI of 20 with VSV(mNeonGreen/P) virus. At 2 h 45 min postinfection, the cells were washed with PBS (catalog number AM9624; Ambion) and fixed with 4% formaldehyde (catalog number 28908; Thermo Fisher) in PBS for 30 min at room temperature (RT). Following permeabilization with 70% ethanol for 2 h at RT, the cells were washed with Stellaris RNA FISH wash buffer A (catalog number SMF-WA1-60; Biosearch Technologies) for 5 min at RT. The cells were probed for N or P plus-or minus-strand RNA with 0.125 M Alexa Fluor 488-or Alexa Fluor 549-labeled probes in Stellaris RNA FISH hybridization buffer (catalog number SMF-HB1-10; Biosearch Technologies) for 16 to 18 h at 37°C. The cells were then washed two times for 30 min at 37°C in Stellaris RNA FISH wash buffer A (catalog number SMF-WA1-60; Biosearch Technologies); the second wash contained Hoechst stain at 1 g/ml. After a 5-min wash with Stellaris RNA FISH wash buffer B (catalog number SMF-WB1-20; Biosearch Technologies), the cells were rinsed three times with PBS and imaged by deconvolution microscopy (DeltaVision OMX SR imaging system). All images were generated by maximum intensity projection using the Z project function in ImageJ (version 2.0.0-rc-59/1.51w). RNA spots were quantified using StarSearch, developed by the Raj laboratory (https://www.seas.upenn.edu/~rajlab/StarSearch/launch.html).
VSV replication assays. Amounts of 1 ϫ 10 4 HT1080 cells were plated in a 96-well plate format, and TRIM69, Mx1, CD68, or RFP expression was induced by overnight treatment with 0.5 g/ml doxycycline hyclate (Sigma-Aldrich). The cells were infected with 30 PFU of VSV(nLuc) per well the next day. At 20 h.p.i. or time points indicated in Fig. 2, supernatant was collected, cells were lysed in passive lysis buffer (catalog number E1941; Promega), and luciferase was measured using the Nano-Glo luciferase assay system (catalog number N1130; Promega) and Modulus II multimode microplate reader (Turner Bio-Systems), or the viral titers in supernatant containing virions were determined on BHK21 cells under a methyl cellulose overlay.
To compare the growth of the TR viruses, 1.2 ϫ 10 6 Vero cells were infected with the different viruses at an MOI of 0.05. Aliquots of the supernatants were harvested at 8, 12, 16, 20, and 24 h.p.i., and the titers determined by cytometry on BSR-T7 cells. Titers are expressed in number of infectious units per ml, i.e., the number of virions leading to detectable expression of eGFP in BSR-T7 cells per ml.
Selection of TRIM69-resistant (TR) viruses. TRIM69-resistant VSV IND (eGFP) and VSV IND (eGFP/P) were selected by plaque assay on HT1080-TagRFP/TRIM69 cells. Cells were seeded and simultaneously treated with 0.5 g/l doxycycline (catalog number D9891; Sigma). Sixteen hours later, cells were infected for 1 h with 1:10 dilutions of the viral stocks and overlaid with medium containing 0.25% agarose. Plaques were picked and amplified once on HT1080 cells expressing TagRFP/TRIM69 and then on BSR-T7 cells.
For the experiment whose results are shown in Fig. 11B, cells were lysed in 50 l of 20 mM Tris-HCl, pH 8, 150 mM NaCl, 0.6% NP-40, 2 mM EDTA, and 1ϫ cOmplete protease inhibitor cocktail (catalog number 4693116001; Roche). Soluble proteins were separated on 10% acrylamide gels, transferred onto nitrocellulose membranes, and incubated with mouse anti-RFP antibody (catalog number ab125244; Abcam), rabbit anti-GFP antibody (catalog number ab6556; Abcam), or mouse antiactin antibody (catalog number A5316; Sigma), followed by incubation with HRP-conjugated anti-mouse antibody (catalog number 31430; Invitrogen) or anti-rabbit antibody (catalog number A0545; Sigma). HRP activity was visualized using the Pierce ECL Western blotting kit (catalog number 32209; Thermo Scientific) and imaged with an Amersham Imager 600 (GE Healthcare). Protein band intensities were quantified using ImageJ software.
Radiolabeling and analysis of virion proteins. For production of virions containing radiolabeled proteins, BSR-T7 cells were seeded in a 150-mm dish in DMEM-10% FBS. The next day, cells were