A Viral RNA Structural Element Alters Host Recognition of Nonself RNA

See allHide authors and affiliations

Science  14 Feb 2014:
Vol. 343, Issue 6172, pp. 783-787
DOI: 10.1126/science.1248465


Although interferon (IFN) signaling induces genes that limit viral infection, many pathogenic viruses overcome this host response. As an example, 2'-O methylation of the 5′ cap of viral RNA subverts mammalian antiviral responses by evading restriction of Ifit1, an IFN-stimulated gene that regulates protein synthesis. However, alphaviruses replicate efficiently in cells expressing Ifit1 even though their genomic RNA has a 5′ cap lacking 2'-O methylation. We show that pathogenic alphaviruses use secondary structural motifs within the 5′ untranslated region (UTR) of their RNA to alter Ifit1 binding and function. Mutations within the 5′-UTR affecting RNA structural elements enabled restriction by or antagonism of Ifit1 in vitro and in vivo. These results identify an evasion mechanism by which viruses use RNA structural motifs to avoid immune restriction.

Keeping Alphaviruses Under Wraps

Viruses mutate to avoid detection, and the host responds in kind. For example, 2′-O methylation of the 5′ cap of viral RNA allows viruses to escape detection by the interferon-stimulated host defense protein, IFIT1. Alphaviruses, however, lack this modification but are able to remain undetected in the presence of IFIT1. How? Using a combination of viral mutants and biochemical analysis, Hyde et al. (p. 783, published online 30 January) found that alphaviruses contain secondary structural motifs in the 5′ untranslated region of their genomic RNA that allow them to avoid detection by IFIT1. When these regions were rendered nonfunctional, IFIT1 was able to keep the virus under control.

Eukaryotic mRNA contains a 5′ cap structure with a methyl group at the N-7 position (cap 0). In higher eukaryotes, methylation also occurs at the 2′-O position of the penultimate and antepenultimate nucleotides to generate cap 1 and 2 structures, respectively. Many viral mRNAs also display cap 1 structures. Because cytoplasmic viruses cannot use host nuclear capping machinery, some have evolved viral methyltransferases for N-7 and 2′-O capping or mechanisms to “steal” the cap from host mRNA (1). Whereas N-7 methylation of mRNA is critical for efficient translation (2), cytoplasmic viruses encoding mutations in their viral 2′-O-methyltransferases are inhibited by IFIT proteins (37), which belong to a family of interferon (IFN)–stimulated genes (ISGs) induced after viral infection [reviewed in (8)]. Thus, 2′-O methylation of host mRNA probably evolved, in part, to distinguish self from nonself RNA (9, 10).

Alphaviruses are positive-strand RNA viruses that replicate in the cytoplasm and lack 2′-O methylation on the 5′ end of their genomic RNA (11, 12) and thus should be restricted by IFIT proteins. To assess the role of IFIT1 in limiting alphavirus replication, we silenced its expression in human HeLa cells and then infected them with Venezuelan equine encephalitis virus (VEEV) strain TC83, an attenuated New World alphavirus. In cells with reduced IFIT1 expression, TC83 replicated to higher levels (Fig. 1A). To determine whether this phenotype occurred in vivo, wild-type (WT) and Ifit1−/− C57BL/6 mice were infected with TC83. In contrast to WT mice, Ifit1−/− mice succumbed to TC83 infection (Fig. 1B) and sustained a higher viral burden (Fig. 1, C and D, and fig. S1), especially in the brain and spinal cord.

Fig. 1 VEEV TC83 but not TRD is restricted by Ifit1.

(A) Flow cytometry contour plots showing infection of TC83 in IFN-β–treated HeLa cells transduced with short hairpin RNA (shRNA) against a scrambled nonsilencing control (NSC), human STAT2, or human IFIT1 (shNSC versus shIFIT1, P < 0.003). One representative experiment of four is shown. This phenotype was confirmed with a second shRNA against IFIT1. (B) Survival of 4-week-old WT mice (n = 10) and Ifit1−/− mice (n = 10) after subcutaneous (s.c.) infection with 106 focus-forming units (FFU) of TC83. Results are pooled from three independent experiments. P values for survival were calculated using the log-rank test. (C and D) Viral burden in 4-week-old WT or Ifit1−/− mice infected s.c. with 106 FFU of TC83, as measured in (C) a draining popliteal lymph node (DLN) and (D) the brain. Results are from five to nine mice per tissue. Asterisks indicate statistically significant differences, as judged by an unpaired t test (**P < 0.005, ***P < 0.0001). Dashed lines indicate the limit of detection of the assay. (E) WT and Ifit1−/− MEFs were pretreated with 10 IU/ml of IFN-β for 12 hours or left untreated, and then infected with TC83 [with a multiplicity of infection (MOI) of 0.1]. Supernatants were harvested for virus titration (WT versus Ifit1−/−, P > 0.2; WT + IFN-β versus Ifit1−/− + IFN, 12, 24, and 36 hours after infection, P < 0.03). Each point represents the average of three experiments performed in triplicate, and error bars represent the standard error of the mean (SEM). P values were determined by an unpaired t test. (F) Survival curves of 8-week-old WT mice (n = 10) and Ifit1−/− mice (n = 24) after s.c. infection with 50 plaque-forming units (PFU) of TRD. Results are pooled from two independent experiments. P values for survival were calculated using the log-rank test.

We next analyzed the growth of TC83 in mouse embryonic fibroblasts (MEFs). Although untreated WT and Ifit1−/− MEFs supported TC83 infection equivalently (Fig. 1E), IFN-β pretreatment preferentially inhibited replication in WT cells. However, an absence of Ifit1 was sufficient to restore infection. A similar trend was observed with Ifit1−/− dendritic cells and cortical neurons (fig. S2, A and B). TC83 infection in Ifit1−/− MEFs remained partially inhibited by IFN-β treatment, indicating that additional ISGs restrict viral replication (1315). The similarity of infection by TC83 in untreated WT and Ifit1−/− MEFs probably reflects the ability of alphaviruses to antagonize the induction of type I IFN and ISGs (16, 17).

TC83 was generated after passage of the virulent Trinidad donkey (TRD) VEEV strain and contains two changes that attenuate virulence (18). One mutation occurs at nucleotide 3 (nt 3, G3A) in the 5′-UTR and increases the sensitivity of TC83 to type I IFN (17). We hypothesized that the 5′-UTR mutation might explain the differential sensitivity to Ifit1 and the pathogenicity of TC83 and TRD. To begin to test this hypothesis, WT and Ifit1−/− mice were infected with TRD (Fig. 1F). WT and Ifit1−/− mice succumbed to TRD infection without differences in survival time or mortality. Thus, in contrast to TC83, TRD was relatively resistant to the antiviral effects of Ifit1.

To determine whether the effect of the G3A mutation was independent of the TC83 structural genes, which contain a second attenuating mutation (19), we assessed replication in WT and Ifit1−/− MEFs of two isogenic chimeric VEEV/Sindbis (SINV) viruses (20); these encode the 5′-UTR and nonstructural proteins of TRD and structural proteins of SINV, and differ only at nt 3 [(G3)VEE/SINV and (A3)VEE/SINV] (fig. S3, A and B). In IFN-β–pretreated WT MEFs, (A3)VEE/SINV was not recovered from culture supernatants (Fig. 2A). However, in IFN-β–treated Ifit1−/− MEFs, (A3)VEE/SINV infection was partially restored. In contrast, (G3)VEE/SINV replicated equivalently in IFN-β–treated WT and Ifit1−/− MEFs (Fig. 2B), indicating that a G at nt 3 renders VEEV resistant to inhibition by Ifit1.

Fig. 2 Mutations in the 5′-UTR determine Ifit1 sensitivity in vitro.

(A and B) Growth kinetics of (A3)VEE/SINV and (G3)VEE/SINV viruses in WT and Ifit1−/− MEFs. Cells were pretreated with 1 IU/ml of IFN-β for 12 hours or left untreated, and then infected with (A3)VEE/SINV or (G3)VEE/SINV (MOI of 0.1). Supernatants were harvested at indicated times for virus titration [(A3)VEE/SINV: WT + IFN-β versus Ifit1−/− + IFN-β, 36 and 48 hours after infection, P < 0.006]. Each point represents the average of three independent experiments performed in triplicate, and error bars represent the SEM. P values were determined using an unpaired t test. Dashed lines indicate the limit of detection of the assay. (C and D) Growth kinetics of (G3)VEE/SINV, (A3U24)VEE/SINV, (A3U24;A20U)VEE/SINV, and (A3U24;20_21insC)VEE/SINV viruses in WT (C) and Ifit1−/− (D) MEFs. Experiments and analysis were performed as in (A). (E to I) Thermal denaturation of A3, G3, A3U24, A3U24;A20U, and A3U24;20_21insC RNA as measured by circular dichroism spectrosopy at 210 nm. RNA was heated from 5° to 95°C at a rate of 1°C/min, and readings were collected every 1°C to monitor unfolding. Data are represented as the change in molar ellipticity as a function of temperature (dθ/dT), and red arrows indicate major maxima. One representative experiment of two is shown.

RNA secondary structure algorithms predicted differences in base pairing at the 5′ end of the UTR of G3 and A3 RNA [fig. S3A and (20, 21)]. The imino region of a two-dimensional nuclear Overhauser effect spectroscopy nuclear magnetic resonance spectrum revealed that A3 RNA displayed less secondary structure and base pairing than G3 RNA (fig. S4, A and B) and fewer cross peaks in the corresponding 1H/15N heteronuclear single-quantum coherence (HSQC) spectrum (fig. S4, C and D). On the basis of these data, we hypothesized that the stable stem-loop structure in the 5′-UTR of TRD compensated for the absence of 2′-O methylation of alphavirus RNA. To determine whether the secondary structure or primary sequence modulated Ifit1 susceptibility, we analyzed the growth of VEE/SINV containing the A3 nt mutation that also had compensatory mutations that were predicted to restore the 5′-UTR stem-loop (Fig. 2, C and D, and fig. S3C). Although two of the mutants tested (A3U24 and A3U24;A20U) showed increased [relative to (A3)VEE/SINV] but limited growth in IFN-β–treated WT MEFs, a third mutant (A3U24;20_21insC) replicated to levels comparable to (G3)VEE/SINV in IFN-β–treated WT and Ifit1−/− MEFs. Mutants that replicated less well in IFN-β–treated WT MEFs (A3U24 and A3U24;A20U) were predicted to have less stable minimum free energy structures relative to (A3U24;20_21insC)VEE/SINV and (G3)VEE/SINV. To further define the requirements in the 5′-UTR for evasion of Ifit1 restriction, we evaluated additional viral mutants: one that changed the sequence of the A3U24 loop but retained the less stable stem structure of the parent A3U24 5′-UTR [(LOOP)VEE/SINV] (21), and two G3 variants with more stable hairpins (G3;C19C20)VEE/SINV that contained additional nucleotide repeats (AUG and AUG2) appended to the 5′ end (fig. S5A). The latter (AUGn)VEE/SINV mutants were relevant because RNA recognition by IFIT proteins reportedly requires a 5′ overhang of 3 to 5 nts (22). Alteration of the loop sequence [(LOOP)VEE/SINV] did not relieve Ifit1-mediated restriction (fig. S5B). However, G3 mutants with an overhang of 3 or more nts at the 5′ end became sensitive to Ifit1-dependent antiviral effects (fig. S5C).

To assess whether nucleotide changes altered the stability of the VEEV 5′-UTR, we monitored RNA unfolding by circular dichroism spectrometry (fig. S6). Changes in ellipticity as a function of temperature were analyzed (Fig. 2, E to I, and table S1); we observed several maxima, presumably corresponding to major cooperative unfolding events (Fig. 2, E to I). We detected more-pronounced maxima near 75°C in all but the A3 RNA, confirming that A3 and G3 RNA have different stabilities. The A3U24;20_21insC mutant RNA displayed the most stable secondary structure. Computational analyses suggested that even closely related RNA sequences (such as A3 and A3U24) have different ensemble free energy and diversity (table S2). Differences in the base pairing probability were noted, which further support structural differences between A3 and G3 RNA (fig. S7). We also measured melting temperature (Tm) values (table S1), which showed an inverse correlation between Ifit1 susceptibility and base-pairing stability. These analyses suggest that G3 and A3U24;20_21insC 5′-UTR RNAs adopt more stable conformations, which correlates with antagonism of Ifit1.

To validate that changes at nt 3 determined sensitivity to Ifit1 independently of other VEEV-encoded factors, we repeated experiments with isogenic variants of TC83 and an enzootic VEEV strain, ZPC-738 (Fig. 3, A to D, and fig. S3D). Whereas TC83 replicated poorly in IFN-β–treated WT MEFs, the isogenic nt 3 mutant TC83 A3G showed increased replication (Fig. 3A), confirming that the A3G mutation confers resistance to type I IFN. However, unlike that seen with (G3)VEE/SINV (Fig. 2B), the phenotype of TC83 A3G in IFN-β–treated WT MEFs did not fully recapitulate the restoration seen in IFN-β–treated Ifit1−/− MEFs (compare Fig. 3A to Fig. 3B), suggesting that additional viral elements may be inhibited by Ifit1. Infection of the mutant ZPC-738 G3A in IFN-β–treated WT MEFs was decreased as compared to WT ZPC-738, whereas the infection of WT and G3A ZPC-738 was equivalent in IFN-β–treated Ifit1−/− MEFs (Fig. 3, C and D).

Fig. 3 Mutations that alter the secondary structure of the 5′-UTR affect pathogenicity in vivo.

(A to D) Growth kinetics of isogenic TC83 WT and A3G [(A) and (B)] or ZPC-738 WT and G3A [(C) and (D)] in WT and Ifit1−/− MEFs. Cells were pretreated with 10 IU/ml of IFN-β for 12 hours (TC83) or 100 IU/ml of IFN-β for 8 hours (ZPC738), or left untreated, and then infected with the respective viruses (MOI of 0.1) [TC83 versus TC83(A3G): WT + IFN-β, 36 and 48 hours after infection, P < 0.006; ZPC738 versus ZPC738(G3A): WT + IFN-β, 24 hours after infection, P < 0.0001]. Each point represents the average of two (ZPC-738) or three (TC83) independent experiments performed in triplicate, and error bars represent the SEM. P values were determined using the unpaired t test. Dashed lines indicate the limit of detection of the assay. (E and F) Survival studies of isogenic ZPC-738 WT and G3A (E) and TC83 WT and A3G (F) viruses in WT and Ifit1−/− mice. Mice were infected s.c. with 101 PFU of ZPC-738 (WT, n = 6; Ifit1−/−, n = 15) or ZPC-738(G3A) (WT, n = 8; Ifit1−/−, n = 15) and 106 PFU of TC83 (WT, n = 18; Ifit1−/−, n = 13) or TC83(A3G) (WT, n = 21; Ifit1−/−, n = 8). ZPC738 versus ZPC738(G3A): WT mice, survival P = 0.0002; mean time to death (MTD) of 5.5 versus 8.3 days, P = 0.0002. ZPC738 versus ZPC738(G3A): Ifit1−/− mice, MTD of 4.0 versus 5.8 days, P < 0.0001. TC83 versus TC83(A3G): WT mice, survival P < 0.0001; TC83 versus TC83(A3G): Ifit1−/− mice, MTD of 8.2 versus 6.3 days, P < 0.003. Experiments were performed twice for ZPC-738 viruses and four times for TC83 viruses. P values for survival were determined as in Fig. 1. P values for MTD were determined using an unpaired t test. (G and H) Growth kinetics of SINV Toto, A5G, and G8U SINV in WT (G) and Ifit1−/− (H) MEFs. Cells were pretreated with 1 IU/ml of IFN-β for 12 hours or left untreated, and then infected with the respective viruses at an MOI of 0.1. SINV Toto versus A5G: WT MEFs + IFN-β, P < 0.05; SINV Toto versus G8U, WT MEFs + IFN-β, P < 0.05. Experiments and analysis were performed as in (A).

To assess whether nt 3 mutation reciprocally affects virulence, we infected WT and Ifit1−/− mice with TC83, ZPC-738, and paired isogenic variants (Fig. 3, E and F). In WT mice, ZPC-738 G3A was attenuated as compared to the WT virus. However, no difference in mortality and only a small difference in survival kinetics were observed in Ifit1−/− mice infected with ZPC-738 WT or G3A. In comparison, we observed increased lethality in WT mice infected with TC83 A3G relative to TC83. We also noted a slight decrease in the survival kinetics of Ifit1−/− mice infected with A3G as compared to TC83 WT, suggesting that the A3G change may have additional effects aside from antagonizing Ifit1 function.

To determine whether structures in the 5′-UTR of other alphaviruses functioned analogously, we introduced mutations at either nt 5 or 8 into SINV (Fig. 3, G and H, and fig. S3E). These mutations were selected because they altered the virulence of SINV in rats (23, 24) and were predicted to change the 5′-UTR secondary structure (fig. S3E). An A–to-G substitution at nt 5 resulted in increased viral replication relative to that of the parental virus in IFN-β–pretreated WT MEFs but not in IFN-β–treated Ifit1−/− MEFs, suggesting that the A5G phenotype was specific to Ifit1. Conversely, a substitution at nt 8 (G8U) resulted in a decrease in replication in IFN-β–treated WT MEFs relative to WT SINV, which was restored to comparable levels in IFN-β–treated Ifit1−/− MEFs. This experiment establishes that mutations within the 5′-UTR of an Old World alphavirus also affect Ifit1 antagonism, suggesting that secondary structure at the 5′-UTR might be a more universal mechanism to circumvent Ifit1-mediated restriction.

IFIT1 binds flavivirus RNA lacking 2′-O methylation and blocks translation and binding of eukaryotic translation initiation factors (6, 7, 25). To determine whether Ifit1 differentially affected translation of alphavirus RNA with different 5′-UTR RNA structures, we transfected type 0 capped WT and G3A mutant translation reporter RNA encoding a luciferase gene fused to nsP1 (fig. S3F) (26) into IFN-β–treated or untreated MEFs (Fig. 4, A to D). In WT MEFs treated with IFN-β (Fig. 4A), G3 RNA exhibited greater translation reporter activity relative to A3 RNA. We also detected greater translation of G3 reporter RNA in untreated WT MEFs (Fig. 4B), suggesting that basal Ifit1 expression in these cells may limit A3 RNA translation. However, we observed a greater increase in A3 reporter RNA translation relative to G3 in Ifit1−/− MEFs that were treated with IFN-β or left untreated (Fig. 4, C and D). The higher level of A3 versus G3 RNA translation in Ifit1−/− MEFs was not unexpected, because (A3)VEE/SINV replicates more efficiently than (G3)VEE/SINV in cells lacking type I IFN induction (20). Although A3 RNA has a translation advantage in cells defective in innate immune responses, the G3 nucleotide confers resistance to Ifit1.

Fig. 4 The nt G3 in the 5′-UTR evades translational inhibition by altering Ifit1-RNA binding.

(A to D) Luciferase assays of A3 and G3 TRD translation reporters. WT and Ifit1−/− MEFs were untreated or treated with 100 IU/ml IFN-β for 8 hours, and then electroporated with in vitro synthesized and type 0–capped reporter RNA. Cell lysates were harvested at the indicated time points and assayed for luciferase activity. Each bar represents the average of four independent experiments performed in triplicate. WT MEFs + IFN-β: G3 versus A3, P < 0.0004; WT MEFs, no treatment: G3 versus A3, P < 0.005; Ifit1−/− MEFs + IFN-β, G3 versus A3, P < 0.05 (30, 60, and 120 min). Error bars represent the SEM. P values were determined using an unpaired t test. (E to G) EMSA of A3 and G3 VEEV 5′-UTR RNA bound to recombinant Ifit1. G3 and A3 VEEV 5′-UTR RNA were synthesized in vitro using T7 polymerase (5′-ppp) and then treated with (E) an N-7 methylguanosine capping reagent (cap 0), (F) an N-7 methylguanosine capping reagent and an exogenous 2′-O methyltranferase (cap 1), or (G) no enzymes (5′-ppp). All RNA was labeled with biotin and competed with 3 μg of homologous unlabeled RNA. Cap 0 and cap 1 RNAs were heated at 95°C; 5′-ppp RNA was heated at 70°C, as no specific binding was observed after heating at 95°C. Binding assays were performed with 1 μg of Ifit1. EMSA data are representative of at least three independent experiments. Arrows indicate specific binding of RNA to Ifit1, whereas asterisks indicate nonspecific binding (not competed with unlabeled RNA). G3 and A3 5′-ppp paired samples were run simultaneously on the same gel and cropped as individual panels for presentation purposes. (H) Quantification of Ifit1-A3/G3 RNA binding by filter-binding assay at 4°C. The fraction bound of A3 cap 0 (black squares) and G3 cap 0 (red squares) was normalized to maximum binding and plotted against Ifit1 concentration. Data from A3 (black) and G3 (red) were fitted using the Hill equation. A3 cap 0 KD = 0.030 ± 0.004 μM; G3 cap 0 KD = 0.091 ± 0.007 μM. One representative experiment of three performed in triplicate is shown.

We hypothesized that alphavirus mutants with different 5′-UTR structural stabilities might interact with Ifit1 in a manner that is less compatible with translation. We used electrophoretic mobility shift assays (EMSAs) (Fig. 4, E to G) to determine whether TRD 5′-UTR RNA containing an A3 or G3 and a type 0 cap differentially interacted with Ifit1 (Fig. 4E). We observed significant binding of Ifit1 to A3 RNA but less binding to G3 RNA, suggesting that the secondary structure of the G3 RNA probably inhibited interaction with Ifit1. This conclusion was supported by dot-blot binding studies, which showed a 2- to 10-fold greater affinity [dissociation constant (KD) ~30 nM] of cap 0 A3 RNA as compared to G3 RNA for Ifit1, depending on the incubation conditions (Fig. 4H and fig. S8). The binding of Ifit1 to cap 0 RNA was specific, as it was competed by excess unlabeled 5′-ppp A3 RNA (fig. S7). Exogenous 2′-O methylation of A3 and G3 RNA, which generates a type 1 cap, resulted in less Ifit1 binding (Fig. 4F), which agrees with flavivirus studies (6, 7). When EMSA experiments were repeated in the absence of capping, TRD 5′-UTR RNA containing an A3 or G3 and a free 5′-ppp differentially and weakly recognized Ifit1 (Fig. 4G), which is consistent with experiments demonstrating that single-stranded RNA, but not double-stranded RNA containing a free 5′-ppp, is bound by IFIT1 (22). Excess A3 5′-ppp RNA compared to G3 5′-ppp RNA preferentially competed for Ifit1 binding to type 0 cap A3 RNA [inhibition constant (Ki ) of 3 and 48 μM for A3 and G3 5′-ppp RNA, respectively; fig. S9]. These results suggest that secondary structure in the context of an uncapped RNA can alter Ifit1 binding and may contribute to why negative-stranded viruses with 5′-ppp genomic RNA and highly structured 5′-UTRs (such as filoviruses) are resistant to type I IFN and Ifit1-mediated control. Our results also establish that Ifit1 has a higher affinity for RNA with a type 0 cap than with a free 5′-ppp moiety.

Alphaviruses use a stable 5′-UTR stem-loop structure to antagonize Ifit1 antiviral activity. Although some IFIT proteins bind 5′-ppp RNA (22, 27), it remains to be determined how Ifit1 differentially recognizes capped RNA that displays or lacks 2′-O methylation and how alphavirus 5′-UTR stem-loop structures affect this. Our experiments suggest that genomic RNA elements can function to evade host cell–intrinsic immunity. Thus, structural elements in viral or virus-associated RNA can bind antiviral proteins irreversibly to block function (28, 29) or attenuate binding of host antiviral proteins. It is intriguing to consider that viral RNA structural elements that antagonize Ifit1 recognition may have become targets for other RNA sensors (such as RIG-I and MDA5). Finally, these results may be relevant to pharmaceutical approaches that use mRNA as therapeutics or vaccine design strategies for attenuating alphaviruses and other viruses.

Supplementary Materials

Materials and Methods

Author Contributions

Figs. S1 to S9

Tables S1 to S4

References (30–46)

References and Notes

  1. Acknowledgments: NIH grants U19 AI083019 (M.S.D.), R01 AI104972 (M.S.D.), R01 AI083383 (W.B.K.), and Training Grant AI049820 (D.W.T.) and the Institute for Human Infections and Immunity, University of Texas Medical Branch (UL1TR000071) supported this work. The authors thank I. Frolov, M. MacDonald, and C. Rice for their generosity with alphavirus reagents and experimental advice. This study made use of the National Magnetic Resonance Facility at Madison, Wisconsin, which is supported by NIH grants P41RR02301 (Biomedical Research Technology Program/National Center for Research Resources) and P41GM66326 (National Institute of General Medical Sciences). The luciferase reporter gene constructs and virulent VEEV strains are subject to material transfer agreements with the University of Pittsburgh and the University of Texas Medical Branch, respectively. Use of the virulent VEEV strain (TRD) requires Biosafety Level 3 facilities and U.S. Department of Agriculture approval. The authors have no financial conflicts to disclose. The data reported in this manuscript are tabulated in the main paper and in the supplementary materials.
View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article