key: cord-0292578-2d0nddrl
authors: Hickson, Sarah E.; Brekke, Eden; Schwerk, Johannes; Saluhke, Indraneel; Zaver, Shivam; Woodward, Joshua; Savan, Ram; Hyde, Jennifer L.
title: Sequence diversity in the 3’ untranslated region of alphavirus modulates IFIT2-dependent restriction in a cell type-dependent manner
date: 2021-12-11
journal: bioRxiv
DOI: 10.1101/2021.12.10.472177
sha: bd3ae74dabc627de0de2684d3bdbdcd12e22b92d
doc_id: 292578
cord_uid: 2d0nddrl

Alphaviruses (family Togaviridae) are a diverse group of positive-sense RNA (+ssRNA) viruses that are transmitted by arthropods and are the causative agent of several significant human and veterinary diseases. Interferon (IFN)-induced proteins with tetratricopeptide repeats (IFITs) are a family of RNA-binding IFN stimulated genes (ISGs) that are highly upregulated following viral infection, and have been identified as potential restrictors of alphaviruses. The mechanism by which IFIT1 restricts RNA viruses is dependent on self and non-self-discrimination of RNA, and alphaviruses evade this recognition via their 5’UTR. However, the role of IFIT2 during alphavirus replication and the mechanism of viral replication inhibition is unclear. In this study, we identify IFIT2 as a restriction factor for Venezuelan equine encephalitis virus (VEEV) and show that IFIT2 binds the 3’ untranslated region (3’UTR) of the virus. We investigated the potential role of variability in the 3’UTR of the virus affecting IFIT2 antiviral activity by studying infection with VEEV. Comparison of recombinant VEEV clones containing 3’UTR sequences derived from epizootic and enzootic isolates exhibited differential sensitivity to IFIT2 restriction in vitro infection studies, suggesting that the alphavirus 3’UTR sequence may function in part to evade IFIT2 restriction. In vitro binding assays demonstrate that IFIT2 binds to the VEEV 3’UTR, however in contrast to previous studies VEEV restriction did not appear to be dependent on the ability of IFIT2 to inhibit translation of viral RNA, suggesting a novel mechanism of IFIT2 restriction. Our study demonstrates that IFIT2 is a restriction factor for alphaviruses and variability in the 3’UTR of VEEV can modulate viral restriction by IFIT2. Ongoing studies are exploring the biological consequences of IFIT2-VEEV RNA interaction in viral pathogenesis and defining sequence and structural features of RNAs that regulate IFIT2 recognition.

that are transmitted by arthropods and are the causative agent of several significant human and 23 veterinary diseases. Interferon (IFN)-induced proteins with tetratricopeptide repeats (IFITs) are a 24 family of RNA-binding IFN stimulated genes (ISGs) that are highly upregulated following viral 25 infection, and have been identified as potential restrictors of alphaviruses. The mechanism by 26 which IFIT1 restricts RNA viruses is dependent on self and non-self-discrimination of RNA, and 27 alphaviruses evade this recognition via their 5'UTR. However, the role of IFIT2 during alphavirus 28 replication and the mechanism of viral replication inhibition is unclear. In this study, we identify 29 IFIT2 as a restriction factor for Venezuelan equine encephalitis virus (VEEV) and show that IFIT2 30 binds the 3' untranslated region (3'UTR) of the virus. We investigated the potential role of 31 variability in the 3'UTR of the virus affecting IFIT2 antiviral activity by studying infection with VEEV. 32

Comparison of recombinant VEEV clones containing 3'UTR sequences derived from epizootic 33 and enzootic isolates exhibited differential sensitivity to IFIT2 restriction in vitro infection studies, 34 suggesting that the alphavirus 3'UTR sequence may function in part to evade IFIT2 restriction. In 35 vitro binding assays demonstrate that IFIT2 binds to the VEEV 3'UTR, however in contrast to 36 previous studies VEEV restriction did not appear to be dependent on the ability of IFIT2 to inhibit 37 translation of viral RNA, suggesting a novel mechanism of IFIT2 restriction. Our study 38 demonstrates that IFIT2 is a restriction factor for alphaviruses and variability in the 3'UTR of VEEV 39 analysis software (Tree Star, Inc.). The initial preliminary shRNA screen was performed in 116 duplicate two times independently, and the fold increase in VEEV positive cells was calculated by 117 comparing data to a non-silencing control shRNA, and z-scores calculated. Based on this data a 118 subset of genes was chosen for further validation based on the following criteria: i) hits with a z-119 score >2, ii) hits for which more than two shRNAs against the same gene exhibited 2-fold or 120 greater VEEV positive cells relative to shNSC; iii). The validation screen was performed as 121 described above using 10 IU/mL of IFN-β stimulation, and was performed in quadruplicate 4 times 122 independently. Cas9 was ligated into the pSBtet-pur backbone using 5' AgeI and 3' NotI restriction sites. The U6 139 promoter was then cloned into this new plasmid using the following primers: U6.F 5'-140 ACTACAGGTACC GAGGG-3', U6.R 5'-TCAGTCCTAGGTCTAGAGC-3', pSBtet-pur-Cas9.F 5'-141 TCAGTCCTAGGTCTAGAGC-3', pSBtet-pur-Cas9.R 5'-142 ATGAAGGTACCACATTTGTAGAGGTTTTACTTGC-3'. U6 was ligated into pSBtet-pur-Cas9 143 using 5' KpnI and 3' AvrII restriction sites. As the new pSBtet-pur-Cas9-U6 plasmid contained an 144 addition BbsI site, this was remove using site directed mutagenesis and the following primers: 145 dBbsI.F 5'-TTGG GAAGAT AATAGCAG-3', dBbsI.R 5'-CTGCTATTATCTTCCCAA-3'. 146 147 6 Sequence-specific gRNA sequences were designed using the Broad Institute Genetic 148

Perturbation Platform gRNA design tool to target mouse Ifit2 (accession # NM_008332). The 149 following primers were used to generate Ifit2 gRNA oligonucleotides which were cloned into 150 transduction, we chose a subset of genes for additional validation using a secondary screen. 316

These targets were chosen based on several criteria, including those with z-scores >2-fold, genes 317 for which several individual shRNA sequences were detected as positive hits, genes encoding 318 proteins predicted to be involved in RNA interactions, and targets which have not previously been 319 explored in the context of alphavirus replication. In total we selected 148 unique shRNAs targeting 320 53 genes to perform the validation screen. As a positive control, we also included a shRNA against 321

STAT2. The validation screen was performed as described for the primary screen, and fold 322 increase in VEEV E2-positive cells (Q2) of the transduced population (Q2+Q3) relative to shNSC 323 was calculated ( Figure 1A and B; Table S5 ). Of the 53 genes, knock down of 24 of these resulted 324 in a >3-fold increase in VEEV positive cells. Knock down of 49 of these genes resulted in a >2-325 fold increase in VEEV positive cells, however only 32 of these genes exhibited a >2-fold increase 326 for multiple unique shRNAs ( Figure 1A) . Notably, we identified several genes and pathways which 327 have previously been implicated in restricting the replication of other alphaviruses, including SNX5 328

[52] and the ISG15 pathway (HERC6, UBE2L6; [53] [54] [55] ). We also identified several IFIT proteins 329 (IFIT1 [9], IFIT2, and IFIT3) as restriction factors for VEEV. 330

We have previously identified IFIT1 as a restriction factor for VEEV and other alphaviruses [9] . 332 As IFIT2 has previously been shown to inhibit replication of several other virus families and protect 333 from lethal neuropathogenesis in vivo [14] [15] [16] [17] [18] [19] , and to be upregulated in the brains of mice 334

following neurovirulent SINV infection [56], we decided to further explore the role of IFIT2 in 335 restriction of VEEV replication. To validate our in vitro findings in an in vivo model of pathogenesis, 336 we infected wild type (WT) and Ifit2 knockout (Ifit2 -/-) mice with VEEV ZPC738 and monitored 337 animals for survival ( Figure 1C ). We observed a modest but significant decrease in the average 338 survival time of Ifit2 -/mice infected with ZPC738 relative to WT mice (p = 0.0138), indicating that 339 in addition to inhibition of replication in vitro, IFIT2 plays some role in VEEV restriction in vivo. 340

The VEEV 3'UTR is a target of IFIT2. Like IFIT1, IFIT2 also possesses RNA binding activity [13] , 342 and has been proposed to exert its antiviral activity via a mechanism dependent on this property. 343

Studies from our lab and others have shown that IFIT1 binds and restricts RNAs containing a 5' 344 7-methylguanisine cap (m7G; cap 0), but not 2'-O-methylated cap structures (cap 1) [9, 11, 12, 345 57, 58] . In contrast, the RNA targets of IFIT2 are poorly defined. Previous studies have 346 demonstrated binding specificity of human IFIT2 for poly (AU) RNA as well as RNAs containing 347 AU-rich elements (AREs), but not polyA, polyU, or GC-rich RNA [13] . However, whether any AU-348 rich sequence is sufficient for recognition and binding by IFIT2, or whether other RNA motifs or 349 surrounding sequences influence binding specificity is unknown. Analogous to cellular mRNAs, 350 many +ssRNA viruses also contain AU-rich regulatory elements in the 3' end of their genomes. 351

As such we speculated that the alphavirus 3' terminus which contains AU-rich sequences is a 352 likely target of IFIT2. To explore this hypothesis, we first performed a sliding window analysis of 353 the VEEV TC83 genome to identify regions with AU-rich sequences ( Figure 2A ). We used a Z-354 score cutoff of 2.32 (representing the 99 th percentile) to identify regions containing significantly 355 high AU content (average = 50.2%) and observed multiple regions in the genome with high AU 356 content, including several genes in open reading frame 1 (ORF1) as well as the 3' UTR. As we 357 predicted, the AU content of the 3'UTR, particularly the very 3' end, was highest amongst all the 358 regions identified (Z-score 2.5-4.2; Figure 2B ). This is consistent with previous studies 359 demonstrating IFIT2 binding specificity for AU-rich sequences [13] and proposed interactions with 360 AU-rich elements (AREs) in cellular RNAs [23] . 361

To determine whether IFIT2 indeed binds the VEEV 3'UTR, we generated recombinant mouse 363 IFIT2 (rIFIT2) and performed Differential Radial Capillary Action of Ligand Assay (DRaCALA) [59] 364 to measure interaction of IFIT2 with VEEV TC83 3'UTR RNA ( Figure 2C ). Analogous to filter 365 binding assays, DRaCALA relies on the propensity of nitrocellulose to bind proteins but not small 366 molecules such as RNA. When protein-ligand mixtures are applied to dry nitrocellulose 367 membranes, protein and protein-bound ligands are immobilized at the site of application, whereas 368 unbound ligand diffuses freely through the membrane via capillary action. We in vitro transcribed 369 and P 32 end-labeled RNA corresponding to the last 200 nucleotides of the TC83 genome that 370 constitutes the 3'UTR and 79 nucleotides of upstream sequence and incubated radiolabeled RNA 371 with serial dilutions of rIFIT2 or bovine serum albumin (BSA) as a negative control. Protein-RNA 372 mixtures were then applied to nitrocellulose and the fraction of bound RNA quantified ( Figure 2D ) 373

[59]. We chose to examine ligand binding using an RNA consisting of additional sequence 374 upstream of the 3'UTR as the nucleotides at the start of the 3'UTR are predicted by RNAfold [60] 375 to participate in base-pairing and therefore would be disrupted by examining binding to the 3'UTR 376 sequence alone. In these assays we observed significantly higher binding of rIFIT2 to TC83 3' 377 RNA (Kd = 0.424 ± 0.051) as compared to BSA which exhibited minimal binding to RNA (Kd = 378

1.433 ± 0.028). Notably, the affinity of rIFIT2 for the VEEV 3'UTR is approximately 10-fold lower 379 than that which we previously observed for IFIT1 and the VEEV 5'UTR [9], suggesting significant 380 differences in affinity of these proteins for their respective RNAs which likely impacts their 381 biological activities. 382

To validate whether the AU content of the 3' RNA is important for determining specificity of IFIT2 384 binding to RNA, we performed additional DRaCALA assays comparing binding of rIFIT2 to 3' 385 RNA, and two additional RNAs with lower AU content ( Figure 2E -G). The first of these 386 corresponds to a region in nsp3 (38% AU), and second a synthetic RNA derived from the plasmid 387 (non-viral sequence) encoding the TC83 infectious clone (34% AU). In these experiments, we 388 synthesized, labeled, and purified ~100 nucleotide long RNAs as described above. Shorter RNAs 389

were used for this experiment, as longer RNAs tended to have a higher AU content that were not synthetic RNAs, which exhibited a 4.5-fold and 2.4-fold increase in Kd relative to 3' RNA 399 respectively (compare blue and purple lines in Figure 2F ). Interestingly, we observed that the 400 binding affinity of IFIT2 for the nsp3 RNA was lowest, despite the fact that the AU content of this 401 RNA is slighter higher than the synthetic RNA. At present it is unknown what the exact 402 determinants of IFIT2-RNA binding are (e.g. AU track length, RNA structure) and we are currently 403 exploring this further. Overall, our binding data shows that IFIT2 preferentially binds AU-rich RNA, 404 and suggests that other factors such as exact primary sequence and secondary structure may 405 also modulate the specificity of binding. 406 407

To determine whether 3'UTR sequences are conserved across different VEEV species, we 409 compared full-length and partial 3'UTR sequences from 136 strains representing four major VEEV 410 subtypes (IAB, IC, ID, and IE) ( Figure 2E ; Figure S1 ). As seen with TC83, we observed several 411 stretches of polyAU sequence throughout the 3'UTR, as is also seen in many +ssRNA viruses. 412 3'UTR sequences from IE subtypes were considerably different from IAB, IC, and ID sequences, 413 consistent with the fact that IE subtypes are more evolutionarily divergent from these other 414 subtypes [61, 62] . However, notably we observed that the 3'UTR sequences of different VEEV 415 strains contained multiple nucleotide variations. Although we did not identify a single mutation or 416 groups of mutations that were exclusively found in viruses belonging to a single subtype (except 417 for the more divergent IE viruses), ( Figure 2E , bottom sequence), several mutations appeared 418 more frequently in viruses belonging to a given subtype. Therefore, we speculated that 3'UTR 419 nucleotide variations present in individual subtypes may lead to alteration in the biological 420 functions of viral 3'UTR RNA, including susceptibility to IFIT2 binding and inhibition of replication. 421

In order to test our hypothesis, we introduced select mutations from IC, ID, and IE subtypes into 423 the 3'UTR of the TC83 infectious clone ( Figure 3A ; Table S6 ). VEEV subtypes exhibit genetic 424 diversity and can be divided into different lineages, with epizootic strains being further classified 425 into clades [1]. We chose several different 3'UTR sequences in order to capture the sequence 426 variation observed, particularly within the ID subtype which encompasses several major lineages 427 and several epizootic clades ( Figure S1 ; Table S6 ). We compared replication kinetics of 3'UTR 428 chimeras in primary murine embryonic fibroblasts (MEF) derived from WT and Ifit2 -/mice ( Figure  429 3B-E). MEF were mock or IFN-β pre-treated for 12 hours to induce IFIT2 expression and infected 430 at a MOI of 0.01. Cell culture supernatant was harvested at 1, 6, 12, 24, 36, and 48 hours post-431 infection (hpi) and infectious virus quantified using focus forming assay (FFA). When we 432 compared replication of WT and mutant viruses in IFN-β pre-treated samples, we observed that 433 replication of TC83 and TC83/IC-3'UTR viruses ( Figure 3B and C) was consistently higher in Ifit2 -434 /cells as compared to WT (>1 log) though not statistically significantly (IAB WT vs KO, p=0.3553), 435 although differences in replication of IC in WT vs KO cells was approaching significance (0.0845). 436

In contrast, we observed identical replication kinetics of TC83/ID-3'UTR and TC83/IE-3'UTR 437 mutant viruses in IFN-stimulated WT and Ifit2 -/cells ( Figure 3D and E), suggesting that these 438 viruses are resistant to the antiviral activities of IFIT2. When comparing replication of all viruses 439 in KO cells pre-treated with IFN, no biological or statistical significance was observed ( Figure 3G ; 440 compare red shaded lines). When we compared replication of all four viruses in WT IFN pre-441 treated cells ( Figure 3G ; compare black and grey shaded lines), we observed significant 442 differences between epizootic and enzootic 3'UTR chimeras (IAB vs ID, p=0.0061; IAB vs IE, 443 p=0.0019; IC vs IE, p=0.0184) except for IC vs ID, although this difference was approaching 444 statistical significance (IC vs ID, p=0.0577). Notably, replication kinetics and viral titers in mock-445 treated samples were near identical for all viruses (compare solid lines in Figure 3B -E), indicating 446 that: i) this phenotype is IFN-dependent and; ii) that the observed IFIT2 phenotype is not due to 447 replication defects or advantages caused by introduction of these mutations. Therefore, we 448 conclude that the differential replication observed between IAB and IC versus ID and IE mutants 449

can be attributed to the specific activities of IFIT2. Of note, TC83/IC-3'UTR and TC83/ID-3'UTR 450 mutant viruses differ only by a single nucleotide (U-to-C mutation at nucleotide 11366; compare 451 Figure 3C and D), suggesting that the differences in IFIT2-mediated restriction could potentially 452 be driven by changes in 3'UTR RNA structure and not primary sequence. This is supported by 453 findings that the primary sequence of the IFIT2 ligand (AU-rich RNA) is somewhat broadly 454 defined. Although higher overall AU content clearly coincides with greater IFIT2 binding ( Figure  455 2F and G), IFIT2 is still able to bind RNAs with variable AU content and length of AU-rich tracts. 456

As such we hypothesize that it would be unlikely that a single nucleotide change would 457 significantly alter the linear IFIT2 recognition motif sufficiently to result in a significant change in 458 IFIT2 binding or IFIT2-dependent viral replication, particularly given that the position of the 459 nucleotide change in IC versus ID does not disrupt the AU-rich tracts that lie downstream of this 460 nucleotide ( Figure 3A ). This is further supported by our previous observations that single 461 nucleotide mutations in the alphavirus 5'UTR is sufficient to alter RNA secondary structure in a 462 manner which prevents binding of RNA by IFIT1 [9] . which replicated similarly in both WT and Ifit2 KO cells, TC83 and TC83/IE-3'UTR mutants 488 exhibited significantly less replication in KO cells relative to WT (P = 0.0283 and P = 0.0382 489 respectively; compare black and red lines in Figure 4B , C, E). We also observed a modest (2-5 490 fold) but significant decrease in replication of TC83/ID-3'UTR virus in Ifit2 KO vs WT cells (P = 491 0.0103). Strikingly, we observed that TC83/ID-3'UTR, which differs from IC by only a single 492 nucleotide, replicated to significantly higher titers than TC83/IC-3'UTR in both WT (20-100 fold 493 increase; P < 0.0001) and Ifit2 KO macrophages (20-80 fold increase; P = 0.004) ( Figure 4D) , 494

suggesting that changes in the virus 3'UTR not only affect VEEV IFIT2-dependency, but also 495 impact VEEV replication in an IFIT2-independent manner. This effect was also cell type-496 dependent, and possibly IFN-dependent, as we did not observe any difference in replication of IC 497 versus ID mutants in WT mock-treated MEF or in IFIT2 deficient cells following IFN pre-treatment 498 Table S6 ) and compared 504 replication of these viruses in WT and Ifit2 KO Raw264.7 cells simultaneously ( Figure 4F-I) . 505

Remarkably, all mutants exhibited near-identical replication to each other as well as the parent ID 506 virus, suggesting that the macrophage-specific function of the 3'UTR is conserved amongst 507 different ID subtypes, despite the presence of these nucleotide variations. Collectively, this data 508 suggests that: i) IFIT2 modulates VEEV replication in a cell type-dependent manner; ii) changes 509 in the VEEV 3'UTR alter the IFIT2-dependency of virus replication; iii) mutations in the virus 3'UTR 510 (IC vs ID) also affects replication of VEEV independent of IFIT2, in a cell type manner; iv) while 511 subtype ID viruses contain numerous SNPs, the 3'UTR-dependent replication phenotype of these 512 viruses is conserved. reporter RNAs appeared to trend slightly higher or lower in WT and Ifit2 -/cells under conditions 535 of mock-or IFN pre-treatment respectively, although this was not statistically significant. 536

As we observed contrasting replication phenotypes in fibroblasts and macrophages, we also 538 performed translation reporter assays in WT and Ifit2 CRISPR KO Raw264.7 ( Figure 5E and F), 539 to determine whether these differences could be attributed to differential IFIT2-dependent 540 translation in these cell types. Assays were performed as for MEF (described above), however 541 unlike MEF, Raw264.7 cells were not pre-treated with IFN-β as these cells basally express IFIT2 542 ( Figure S2 ). When we compared translation of reporter RNAs in these cells we observed a modest 543 (~2-4 fold) but significant decrease in translation of all reporter RNAs in WT vs KO cells, consistent 544 with regulation role for IFIT2 in global translation regulation. However, we observed no difference 545 in translation between reporter RNAs in either WT or KO cells, with the exception of repTC83/IE-546 3'UTR, suggesting that 3'UTR-dependent differences in translation do not account for our 547 observed replication phenotypes in either fibroblasts or macrophages. Interestingly, we observed 548 up to a 5-fold (WT) or 7.7-fold (Ifit2 -/-) increase in translation of repTC83/IE-3'UTR relative to other 549 reporters, specifically in macrophages but not fibroblasts. Of note, we observed no difference in 550 translation of either IC or ID in either cell type, despite the fact that the single nucleotide difference 551 between these RNAs accounts for significant differences in replication in both cell lines (Figure 3  552 and 4). Collectively, our translation reporter data suggests that IFIT2 plays a modest role in global 553 regulation of cellular translation in fibroblasts and macrophages, consistent with previous reports 554 of the role of IFIT2 in translation inhibition and regulation [22, 67] . However, the mechanism by 555 which IFIT2 inhibits (MEF) or promotes (Raw264.7) replication of 3'UTR mutants appears to be 556 independent of viral RNA translation. Given that IFIT2 has been shown to bind to and regulate 557 translation of cellular mRNAs [23, 37] , it is possible that IFIT2-dependent translation inhibition of 558 a subset of cellular transcripts could explain these observations. Further studies are necessary to 559 elucidate the molecular mechanism by which IFIT2 regulates replication of VEEV and how the 560 viral 3'UTR contributes to replication, innate immune evasion, and pathogenesis. In this study, we used a shRNA screening approach to identify VEEV restriction factors and 565 identified IFIT2 as a novel viral restriction factor of VEEV replication and pathogenesis. We 566 identified that IFIT2 specifically binds to the 3'UTR of the VEEV genome, which contains AU rich 567 elements, a broadly defined motif abundant in both cellular and +ssRNA viral RNA. This is 568 consistent with previous studies of human IFIT2 which has been shown to bind AU rich RNA 569

sequences from different strains of VEEV, we observed tremendous diversity in their sequences. 571

We also noted that the 3'UTRs of viruses belonging to different subtypes (IAB, IC, ID, and IE) of 572 VEEV were distinct, and speculated that these sequence differences may confer functional 573 changes in 3'UTR RNA that affect replication and innate immune evasion. To test this, we 574 mouse infection studies with VEEV ZPC738 shows that IFIT2 predominantly plays an antiviral 585 role during VEEV pathogenesis ( Figure 1C ). As TC83 is highly attenuated in C57BL/6J mice 586 compared to VEEV ZPC738, such direct comparisons might not be possible at this time. TC83 587 infection causes only modest weight loss in WT C57BL/6J mice, and only minor or modest clinical 588 manifestations, thus this model is not sufficiently robust to distinguish and dissect modest in vivo 589

phenotypes. As such, we are generating 3'UTR chimeras on in other VEEV backgrounds 590 (ZPC738) which will be used to investigate this hypothesis in following studies. We also cannot 591 exclude the possibility that other VEEV sequences outside of the 3'UTR may contribute to IFIT2-592 dependent phenotypes. Nonetheless, our data shows that the biological function of IFIT2 during 593 virus replication differs depending on the cell type, and likely the virus as well. 594 595 Secondly, we observed that changes in the VEEV 3'UTR had a significant effect on virus 596 replication in an IFIT2-dependent manner. While the effect of coding sequence mutations has 597 previously been studied in the context of epizootic VEEV emergence, the role of sequence 598 variability in the 3'UTR has not been investigated. Despite observing opposing IFIT2 phenotypes 599 in fibroblasts and macrophages, changes in the VEEV 3'UTR affected the dependency of VEEV 600 replication on IFIT2 in both cell types. Interestingly, the effect of some 3'UTR mutations on IFIT2-601 dependent phenotypes differed between these cells. For example, in MEF we observed an IFIT2-602 dependent replication phenotype with TC83/IC-3'UTR ( Figure 3C ), but in macrophages we 603 observed no significant difference in replication of this mutant in WT vs KO cells (compare with 604 Figure 4C ). These data may suggest that other cellular factors which are differentially expressed 605 in these cell types also impacts IFIT2. Other IFIT proteins have been demonstrated to form hetero-606 oligomers to modulate their functions. Recent studies have shown that IFIT2 and IFIT3 interact 607 with IFIT1 leading to enhanced cap binding activities of this protein [21, 68] . We propose other 608 host factors which interact with IFIT2 alter the biological properties of IFIT2, and/or IFIT2 609 competes with other RNA-binding proteins for binding to the VEEV 3'UTR and the outcome of 610 this competition is influenced by the underlying sequence and secondary structure of the viral 611

We were particularly intrigued by differences in replication of IC and ID 3'UTR mutant viruses. In 614 MEF we observed that the ID mutant was resistant to IFIT2-mediated inhibition (compare Figure  615 3C and 3D), and in macrophages observed a significant increase in replication (~20-200 fold) 616 relative to the IC mutant (compare Figure 4C for conferring this epizootic phenotype, which is characterized by increased type I IFN resistance 624 [7, 8, 72, 73] , as well as high titer viremia and pathogenesis in equines which serve as 625 amplification hosts during epizootic episodes [7] . While these mutations are critical for emergence 626 of epizootic VEEV, comparative analysis of epizootic and enzootic sequences reveal that 627 epizootic viruses also acquire other mutations elsewhere in the genome, including the 3'UTR 628 ( Figure S1 ). Significantly, a high proportion of these mutations are synonymous (unpublished 629 data). While many of these mutations can likely be explained by divergent evolution due to 630 geological constraints, we speculate that some mutations may confer biological properties (e.g. 631 type I IFN resistance) that, although not essential, may aid in emergence of epizootic subtypes. 632

Thus we speculate that acquisition of mutations that alter the function of underlying diversity in for IFIT2 in global RNA translation. However, no significant differences were observed between 643 mutant 3'UTR reporter RNAs in either WT or KO cells. Interestingly, we observed a significant 644 increase in translation of IE 3'UTR reporter RNA relative to the other VEEV 3'UTRs, specifically 645 in macrophages. These data would suggest that mutations in the IE 3'UTR enhances viral RNA 646 translation specifically in macrophages. This again bolsters our hypothesis that cell-type specific 647 host factors expressed in macrophages are critical for determining VEEV replication, however 648 how this enhanced translation would impact replication and pathogenesis of IE subtypes in vivo 649 is at present unclear. Nonetheless, these data show that our observed IFIT2-dependent 650 replication phenotypes of 3'UTR mutants cannot be explained by differences in translation alone. 651

In addition to demonstrating that the VEEV 3'UTR modulates IFIT2-dependent activities, we also 653 observed IFIT2-independent roles for the 3'UTR in VEEV replication and RNA translation. In summary, we identified a role for IFIT2 in restricting VEEV replication and pathogenesis in vitro 665 in vivo. We demonstrated that IFIT2 targets VEEV by binding to the 3'UTR and show that changes 666 in the VEEV 3'UTR sequence modulate the ability of IFIT2 to inhibit VEEV replication in a cell 667 type-dependent manner. In contrast to previous studies, our data suggests that IFIT2 affects 668 replication of VEEV 3'UTR mutants via a mechanism independent of translation. We also 669 demonstrated an IFIT2-independent role for the VEEV 3'UTR in replication in macrophages. We thank the members of the Ram laboratory for helpful scientific discussions and technical 789 assistance. We thank Dr. Michael Diamond for generously providing reagents for this work, and 790

we thank Jason Smith for scientific discussion. 791

The authors declare that they have no conflict of interest. 792 793 

Venezuelan equine encephalitis

Resistance to alpha/beta interferons correlates with the epizootic and virulence 797 potential of Venezuelan equine encephalitis viruses and is determined by the 5' noncoding region 798 and glycoproteins

Mutations in the E2 glycoprotein of Venezuelan 800 equine encephalitis virus confer heparan sulfate interaction, low morbidity, and rapid clearance 801 from blood of mice

Positively charged amino acid substitutions in the e2 envelope glycoprotein are 803 associated with the emergence of venezuelan equine encephalitis virus

Vector infection determinants of Venezuelan equine 806 encephalitis virus reside within the E2 envelope glycoprotein

Attenuation of Venezuelan equine encephalitis virus strain TC-83 is encoded 808 by the 5'-noncoding region and the E2 envelope glycoprotein

Envelope glycoprotein mutations mediate equine amplification and virulence 810 of epizootic venezuelan equine encephalitis virus

Venezuelan equine encephalitis virus in the guinea pig model: evidence for 812 epizootic virulence determinants outside the E2 envelope glycoprotein gene

A viral RNA structural element alters host recognition of nonself RNA

Sequestration by IFIT1 impairs translation of 2'O-unmethylated capped RNA

2'-O methylation of the viral mRNA cap by West Nile virus evades ifit1-819 dependent and -independent mechanisms of host restriction in vivo

Ifit1 inhibits Japanese encephalitis virus replication through binding to 5' capped 822 2'-O unmethylated RNA

Crystal structure of ISG54 reveals a novel RNA binding structure and potential 824 functional mechanisms

A role for Ifit2 in restricting West Nile virus infection in the brain

Ifit2 deficiency results in uncontrolled neurotropic coronavirus replication and 830 enhanced encephalitis via impaired alpha/beta interferon induction in macrophages

Interferon-induced protein Ifit2 protects mice from infection 833 of the peripheral nervous system by vesicular stomatitis virus

Sendai virus pathogenesis in mice is prevented by Ifit2 and 835 exacerbated by interferon

Ifit2 Is a Restriction Factor in Rabies Virus Pathogenicity

Identification of Japanese encephalitis virus-inducible genes in mouse brain and 838 characterization of GARG39/IFIT2 as a microtubule-associated protein

IFIT3 and IFIT2/3 promote IFIT1-mediated translation inhibition by enhancing 841 binding to non-self RNA

Induction and mode of action of the viral stress-inducible murine 843 proteins, P56 and P54. Virology

Forced IFIT-2 expression represses LPS induced TNF-alpha expression at 845 posttranscriptional levels

The interferon stimulated gene 54 promotes apoptosis

MicroRNA-645, up-regulated in human adencarcinoma of gastric esophageal 849 junction, inhibits apoptosis by targeting tumor suppressor IFIT2

Long non-coding RNA LINC00161 sensitises osteosarcoma cells to cisplatin-851 induced apoptosis by regulating the miR-645-IFIT2 axis

Inhibition of Proteasome Activity Induces Aggregation of IFIT2 in the Centrosome 853 and Enhances IFIT2-Induced Cell Apoptosis

Curcumin induces apoptosis in human leukemic cell lines through an IFIT2-855 dependent pathway

Depleting IFIT2 mediates atypical PKC signaling to enhance the migration and 857 metastatic activity of oral squamous cell carcinoma cells

Blocking TNF-alpha inhibits angiogenesis and growth of IFIT2-depleted metastatic 859 oral squamous cell carcinoma cells

The LIM protein AJUBA promotes colorectal cancer cell survival through suppression 861 of JAK1/STAT1/IFIT2 network

Decreased IFIT2 Expression Promotes Gastric Cancer Progression and Predicts Poor 863 Prognosis of the Patients

PLZF inhibits proliferation and metastasis of gallbladder cancer by regulating IFIT2

IFIT2 is an effector protein of type I IFN-mediated amplification of 867 lipopolysaccharide (LPS)-induced TNF-alpha secretion and LPS-induced endotoxin shock

ISG54 and ISG56 are induced by TLR3 signaling in U373MG human astrocytoma 870 cells: possible involvement in CXCL10 expression

Interferon-stimulated gene (ISG) 60, as well as ISG56 and ISG54, positively 872 regulates TLR3/IFN-beta/STAT1 axis in U373MG human astrocytoma cells

Influenza virus repurposes the antiviral protein IFIT2 to promote translation of viral 875 mRNAs

Antigenic analysis of the surface glycoproteins of a 877

Venezuelan equine encephalomyelitis virus (TC-83) using monoclonal antibodies

Genome engineering using the CRISPR-Cas9 system

Optimized Sleeping Beauty transposons rapidly 882 generate stable transgenic cell lines

The lectin pathway of complement activation contributes to protection from West 884 Nile virus infection

2'-O methylation of the viral mRNA cap evades host restriction by IFIT family 886 members

Alpha/beta interferon inhibits cap-dependent translation of viral but not 888 cellular mRNA by a PKR-independent mechanism

Eastern and Venezuelan equine encephalitis viruses differ in their ability to 890 infect dendritic cells and macrophages: impact of altered cell tropism on pathogenesis

tRNA binding, structure, and localization of the human interferon-induced 893 protein IFIT5

PKR acts early in infection to suppress Semliki Forest virus production and strongly 895 enhances the type I interferon response

PKR-dependent and -independent mechanisms are involved in translational 897 shutoff during Sindbis virus infection

Effects of PKR/RNase L-dependent and alternative antiviral pathways on 899 alphavirus replication and pathogenesis

Interferon Regulatory Factor 1 Protects against Chikungunya Virus-Induced 901 Immunopathology by Restricting Infection in Muscle Cells

Activation of the type I interferon pathway is enhanced in response to human 903 neuronal differentiation

Macromolecular Synthesis Shutoff Resistance by Myeloid Cells Is Critical

Dependent Systemic Interferon Alpha/Beta Induction after Alphavirus Infection

Comparative Characterization of the Sindbis Virus Proteome from 908

Mammalian and Invertebrate Hosts Identifies nsP2 as a Component of the Virion and Sorting Nexin 909 5 as a Significant Host Factor for Alphavirus Replication

ISG15 is critical in the control of Chikungunya virus infection independent of 911 UbE1L mediated conjugation

Identification and characterization of interferon-induced proteins that inhibit 913 alphavirus replication

TRIM25 Enhances the Antiviral Action of Zinc-Finger Antiviral Protein (ZAP)

Identification of genes involved in the host response to neurovirulent alphavirus 917 infection

IFIT1 is an antiviral protein that recognizes 5'-triphosphate RNA

Structure of human IFIT1 with capped RNA reveals adaptable mRNA binding 921 and mechanisms for sensing N1 and N2 ribose 2'-O methylations

Differential radial capillary action of ligand assay for high-throughput 924 detection of protein-metabolite interactions

Using RNAFOLD to predict the activity of small catalytic RNAs

Complete sequence of Venezuelan equine encephalitis 928 virus subtype IE reveals conserved and hypervariable domains within the C terminus of nsP3

Geographic distribution of Venezuelan equine encephalitis virus subtype IE 931 genotypes in Central America and Mexico

A single-site mutant and revertants arising in vivo define early steps in the 933 pathogenesis of Venezuelan equine encephalitis virus

Role of dendritic cell targeting in Venezuelan equine 935 encephalitis virus pathogenesis

Host translation shutoff mediated by non-structural protein 2 is a critical factor 937 in the antiviral state resistance of Venezuelan equine encephalitis virus

Characteristics of alpha/beta interferon induction after infection of murine 940 fibroblasts with wild-type and mutant alphaviruses

Distinct induction patterns and functions of two closely related interferon-942 inducible human genes, ISG54 and ISG56

Human IFIT3 Modulates IFIT1 RNA Binding Specificity and Protein Stability. 944 Immunity

Equine amplification and virulence of subtype IE Venezuelan equine 946 encephalitis viruses isolated during the 1993 and 1996 Mexican epizootics. Emerg Infect Dis

Association of Venezuelan equine encephalitis virus subtype IE with two 949 equine epizootics in Mexico

Genetic evidence that epizootic Venezuelan equine encephalitis (VEE) viruses 951 may have evolved from enzootic VEE subtype I-D virus

Repeated emergence of epidemic/epizootic Venezuelan equine encephalitis 953 from a single genotype of enzootic subtype ID virus

Genetic and phenotypic changes accompanying the emergence of epizootic 955 subtype IC Venezuelan equine encephalitis viruses from an enzootic subtype ID progenitor

Virulence and viremia characteristics of 1992 epizootic subtype IC Venezuelan 958 equine encephalitis viruses and closely related enzootic subtype ID strains

Venezuelan equine encephalitis complex and identification of the source of epizootic viruses

RNA viruses can hijack vertebrate microRNAs to suppress innate immunity. 964 Nature