characterization of the rna silencing suppression activity of the

Characterization of the RNA Silencing Suppression Activity of theEbola Virus VP35 Protein in Plants and Mammalian Cells

Yali Zhu,a Nil Celebi Cherukuri,a Jamie N. Jackel,b,d Zetang Wu,c* Monica Crary,b Kenneth J. Buckley,b David M. Bisaro,b,c,d andDeborah S. Parrisa,b,c

Department of Molecular Virology, Immunology and Medical Genetics,a Department of Molecular Genetics and Plant Biotechnology Center,b Graduate Program inMolecular, Cellular, and Developmental Biology,c and Center for RNA Biology,d The Ohio State University, Columbus, Ohio, USA

Ebola virus (EBOV) causes a lethal hemorrhagic fever for which there is no approved effective treatment or prevention strategy.EBOV VP35 is a virulence factor that blocks innate antiviral host responses, including the induction of and response to alpha/beta interferon. VP35 is also an RNA silencing suppressor (RSS). By inhibiting microRNA-directed silencing, mammalian virusRSSs have the capacity to alter the cellular environment to benefit replication. A reporter gene containing specific microRNAtarget sequences was used to demonstrate that prior expression of wild-type VP35 was able to block establishment of microRNAsilencing in mammalian cells. In addition, wild-type VP35 C-terminal domain (CTD) protein fusions were shown to bind smallinterfering RNA (siRNA). Analysis of mutant proteins demonstrated that reporter activity in RSS assays did not correlate withtheir ability to antagonize double-stranded RNA (dsRNA)-activated protein kinase R (PKR) or bind siRNA. The results suggestthat enhanced reporter activity in the presence of VP35 is a composite of nonspecific translational enhancement and silencingsuppression. Moreover, most of the specific RSS activity in mammalian cells is RNA binding independent, consistent withVP35’s proposed role in sequestering one or more silencing complex proteins. To examine RSS activity in a system without inter-feron, VP35 was tested in well-characterized plant silencing suppression assays. VP35 was shown to possess potent plant RSSactivity, and the activities of mutant proteins correlated strongly, but not exclusively, with RNA binding ability. The results sug-gest the importance of VP35-protein interactions in blocking silencing in a system (mammalian) that cannot amplify dsRNA.

Ebola virus (EBOV), a member of the Filoviridae, is a class Apriority pathogen. The most pathogenic strains (e.g., Zaire)

cause severe hemorrhagic fever in humans, with a fatality rate ofup to 80 to 90% (20). Although there have been several promisingvaccine approaches and postexposure antisense therapeutics innonhuman primates, no licensed vaccines or specific therapeuticscurrently exist for the prevention or treatment of human EBOVinfections (14, 27, 41, 47, 50). Moreover, EBOV blocks the induc-tion of interferon (IFN) and does not respond to exogenous IFNin vivo or in vitro (1, 17, 18, 22, 24). Most of this antagonism ismediated by EBOV VP35 (1–3), a multifunctional 340-amino-acid (aa) protein that is also an essential viral RNA polymerasecofactor and a structural component of the virion (21, 36, 37). Inaddition to its ability to bind double-stranded RNA (dsRNA),EBOV VP35 blocks activation of IRF-3 and/or IRF-7. It also di-rectly or indirectly interacts with a number of cellular proteins,including the kinases I�B kinase � (IKK�) and TBK-1, required tophosphorylate and translocate these IRFs to the nucleus to inducetranscription of type 1 IFNs (2, 7, 8, 38). VP35 also facilitates theaddition of SUMO to IRF-7, most likely by interacting with severalproteins required for the process (8). The C-terminal domain(CTD) of VP35 (aa 215 to 340) is responsible for all, or nearly all,of these activities and has been referred to as the IFN-inhibitorydomain (VP35 IID) (1, 19). VP35 has also been shown to inhibitand reverse the activation of the dsRNA-activated protein kinase R(PKR), another important effector of the IFN pathway. Thoughthe mechanism is unknown, VP35’s effect on PKR is apparentlyRNA independent and prevents the phosphorylation and inacti-vation of the important translation factor eIF-2� (12, 45).

In addition to its effects on the IFN pathway, VP35 has beendemonstrated to be a potent RNA silencing suppressor (RSS) (15).RNA silencing pathways are highly conserved among plants, ani-

mals, fungi, and fission yeast (44) and, therefore, likely representsome of the most primordial defense mechanisms. Indeed, it iswell established that RNA silencing is an innate antiviral defense inplants, and virtually all plant viruses encode one or more RSSproteins that act as pathogenicity determinants (reviewed in ref-erences 10, 31, and 40). Many of these RSSs have been shown toblock small interfering RNAs (siRNAs) and/or pathways requiredfor their generation (4, 28). Moreover, many plant virus RSS pro-teins also interfere with microRNA-directed silencing (25). Therole of RSSs in the pathogenicity of mammalian viruses has beenthe subject of great debate. However, because the microRNApathway is a major posttranscriptional regulatory mechanism inmammals, the ability of a virus to suppress microRNA-directedsilencing globally or specifically could alter the cellular environ-ment to enhance virus replication and/or spread. Thus, the abilityof VP35 to interfere with microRNA-directed silencing could en-hance the ability of EBOV to replicate. The loss of most of the RSSactivity of VP35 with mutation of arginine 312 to alanine (R312A)suggested to earlier investigators that silencing suppression activ-ity required the ability of VP35 to bind dsRNA (15), since theR312A mutant protein failed to bind dsRNA in vitro (7). However,

Received 20 July 2011 Accepted 20 December 2011

Published ahead of print 11 January 2012

Address correspondence to Deborah S. Parris, [email protected].

* Present address: Department of Neurology, University of Michigan, Ann Arbor,Michigan, USA.

Y. Zhu and N. C. Celebi Cherukuri contributed equally to this article.

Copyright © 2012, American Society for Microbiology. All Rights Reserved.

doi:10.1128/JVI.05741-11

3038 jvi.asm.org 0022-538X/12/$12.00 Journal of Virology p. 3038–3049

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://dx.doi.org/10.1128/JVI.05741-11

http://jvi.asm.org

structural studies have revealed that R312 of VP35 is involved notonly in interactions with the backbone phosphates of dsRNA, butalso in protein-protein interactions at the interface of monomersin the asymmetric dimer of the VP35 CTD that forms in cocrystalswith dsRNA (26, 30). Thus, it remains unclear whether RNA si-lencing inhibition by VP35 depends on dsRNA binding.

To better understand the means by which VP35 suppressesRNA silencing and to investigate its ability to interfere withmicroRNA silencing, we employed assays that reflect themicroRNA-directed silencing pathway that is used by mammals,as well as the siRNA-directed pathway involved in natural antivi-ral defenses in plants. Plant silencing suppression assays have theadditional advantage of avoiding any direct or indirect links withIFN pathways. We also report on the ability of the VP35 CTD tobind directly to the small dsRNAs that are mediators of siRNA-and microRNA-directed responses. The results suggest thatmicroRNA-directed silencing suppression in mammalian cells in-volves mostly RNA-independent mechanisms. In contrast, inplants that rely heavily on dsRNA amplification, RNA-dependentmechanisms predominate.

MATERIALS AND METHODSPlasmids. The pCMV-Luc-miR30 reporter and pSuper-miR30 plasmidswere obtained from Bryan Cullen and have been described in detail else-where (54, 56, 57). pCMV-Luc-miR30 allows high-level constitutive ex-pression of an mRNA containing the firefly luciferase (Fluc) open readingframe (ORF) with 8 tandem repeats of a microRNA 30 (miR30) targetsequence in the 3= untranslated region (UTR). The pSuper-miR21 plas-mid was derived by cloning human miR21 sequences into the pSuperbackbone. pSuper-miRs are expressed from the human H1 RNA poly-merase III promoter sequences. pRLCMV was obtained from Promega(Madison, WI) and expresses Renilla luciferase (Rluc) from the humancytomegalovirus (HCMV) RNA polymerase II promoter. The plasmidpEGFP-C2 was obtained from Clontech (Mountain View, CA) for HCMVpromoter-driven expression of enhanced green fluorescent protein(EGFP) in mammalian cells. Plasmid p19-FLAG expresses the plant virus-silencing suppressor under HCMV promoter control (39) and was thekind gift of Kathy Boris-Lawrie (Ohio State University). The EBOV VP35gene (Zaire) was the kind gift of Chris Basler (Mt. Sinai Medical Center).The plasmids pcDNA4/TO, pCEP4-tetR, and pcDNA4/TO-LacZ havebeen described previously (52, 53) and were kind gifts of Feng Yao (Har-vard Medical School). To facilitate detection, the VP35 ORF was placeddownstream of an N-terminal c-myc epitope tag composed of the DNAsequence 5=-ATGGAACAAAAACTCATCTCAGAAGAGGATCTG-3=and cloned into the pcDNA 3.1 and pcDNA4/TO mammalian expressionvectors. The pcDNA4/TO vector contains the HCMV immediate-earlypromoter and 2 tetracycline (TET) operator sequences to permit doxycy-cline (DOX)-inducible expression in cells that express the TET repressor(52) but constitutive expression in cells without the TET repressor. Toobtain maltose binding protein (MBP) fusions with the CTD of VP35, theplasmid pMAL-5cE (New England Biolabs, Beverly, MA) was cleaved withKpnI, blunted, and ligated to the sequences encoding the myc tag linked toaa 215 to 340 of the wild-type VP35 gene (obtained by linked PCR). TheK309A, R312A, and G333S mutations in the VP35 gene were prepared bynested PCR and transferred to the indicated expression plasmids byswitching subdomains containing these mutations with the analogouswild-type VP35 sequences. Plasmids containing the p19 plant virus-silencing suppressor and �-glucuronidase (GUS) genes downstream ofthe constitutive Cauliflower mosaic virus 35S promoter were positive andnegative controls, respectively, for plant silencing suppression assaysand have been described elsewhere (49). The myc-VP35 genes (wild typeand mutants) were also cloned downstream of the 35S promoter in thesame plasmid background for expression in plants. DNA sequencing was

performed to confirm the open reading frames within all plasmids con-structed. Additional cloning details are available upon request.

Transient transfections. Baby hamster kidney (BHK), human 293T,or 293T derivative cells were seeded into 35- or 60-mm culture dishesin antibiotic-free Dulbecco’s modified minimum essential medium(DMEM) containing 10% fetal bovine serum and incubated overnight at34°C in a humidified 5% CO2 atmosphere to achieve 50 to 80% conflu-ence. Approximately 1 h prior to transfection, the culture medium waschanged to serum- and antibiotic-free DMEM containing 0.1 mM non-essential amino acids. DNA mixtures containing constant amounts ofDNA (700 ng/35-mm dish or 2 �g/60-mm dish) and lipofectamine 2000(Invitrogen; 4.25 �l/dish) were added to cells in Opti-MEM low-serummedium as described previously (51). The DNA mixtures were preparedwith the amounts of individual plasmids indicated for each experiment,with the remaining DNA comprised of empty-vector sequences (pUC19or pCDNA3.1). Four hours after transfection, the medium was replacedwith DMEM containing 10% fetal bovine serum, and the cells were incu-bated as before and were harvested as indicated for each experiment. Forsome experiments, as indicated, a second transfection was performed 24 hafter the first.

Preparation and cultivation of stable cell lines. Human 293T cellswere transfected with pCEP4-tetR plasmid (700 ng) and incubated for 2 to4 weeks in medium containing hygromycin (50 �g/ml) to obtain TETrepressor (tetR)-expressing cell lines. Culture medium was replenishedevery 3 to 4 days, and clones resistant to hygromycin were picked andexpanded. The clones were functionally screened for the presence of tetRprotein by the ability to regulate expression of a test plasmid (pcDNA4/TO-LacZ) containing the Escherichia coli lacZ gene downstream of theTET operator. When transfected with up to 20 ng of pcDNA4/TO-LacZand in the absence of induction, the tetR-expressing cell line selected (6c3)displayed no detectable �-galactosidase (�-Gal) protein by either X-Gal(5-bromo-4-chloro-3-indolyl-�-D-galactopyranoside) staining or immu-noblotting with �-Gal antibody. However, high levels of �-Gal expressionwere observed when similarly transfected cells were incubated in the pres-ence of DOX (1 �g/ml) for 24 h. Cell line 6c3 was transfected with thezeocin-selectable vector pcDNA4/TO containing either the VP35 (wildtype or mutant) or lacZ gene, and clones resistant to zeocin (250 �g/ml)were isolated. Stably transfected cell lines were maintained continuouslyin medium containing hygromycin (50 �g/ml) and zeocin (250 �g/ml)and were screened by immunoblotting for relevant protein expression inthe absence or presence of DOX (1 �g/ml) for 24 h. The C-terminal halvesof the VP35 genes from mutant and wild-type cell clones were PCR am-plified from purified total cellular DNA, and the sequences were con-firmed.

Antibodies and immunoblotting. Cells from parallel cultures usedfor luciferase assays were harvested by scraping the cell monolayers intothe medium, followed by low-speed centrifugation. The cell pellets werewashed twice in ice-cold phosphate-buffered saline (PBS), flash frozen inliquid nitrogen, and stored at �80°C. The cells were suspended in buffercontaining 0.1 M Tris-Cl, pH 8.0, 150 mM NaCl, 10% glycerol, 0.5%Nonidet P-40, 0.5% Na deoxycholate and lysed by ultrasonic disruption at0°C. The suspensions were clarified by low-speed centrifugation, and theprotein content in the supernatant was determined using a Coomassieblue dye-based assay (Bio-Rad) with bovine serum albumin as the stan-dard. Equivalent amounts of protein were separated by SDS-PAGE, andthe proteins were transferred to nitrocellulose membranes. The mem-branes were blocked in PBS containing 10% nonfat dry milk, washedextensively with PBS containing 0.1% Tween 20, and then incubated witha 1:5,000 dilution of mouse monoclonal antibody 46-0603 to c-myc (In-vitrogen), a 1:666 dilution of rabbit polyclonal antibody to �-Gal (5=-3=Inc., West Chester, PA), or a 1:500 dilution of rabbit polyclonal antibodyto GFP (Santa Cruz), as indicated. Following extensive washing, the blotswere incubated with a 1:10,000 dilution of horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit IgG (Thermo) and developedusing the Super Signal West Pico Chemiluminescent kit according to the

Ebola Virus VP35 RNA Silencing Suppression Activities

March 2012 Volume 86 Number 6 jvi.asm.org 3039

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

instructions of the manufacturer. The blots were exposed to X-ray film for1 to 10 s and developed.

Luciferase assays. The dual-luciferase reporter assay kit from Pro-mega (Madison, WI) was employed, using an active lysis procedure ac-cording to the instructions of the manufacturer. Cells in 35- or 60-mmdishes were seeded and/or pretreated as indicated for each experiment.Twenty-four hours later, the cells were transfected with pCMV-Luc-miR30 (30 ng/35-mm dish or 90 ng/60-mm dish), pCMV-Rluc (2 or 6 ngfor 35- or 60-mm dishes, respectively), mixtures of pSuper-miR30 and/orpSuper-miR21 as indicated in each experiment, and vector plasmid filleras described above. In some experiments, test expression plasmids weretransfected 24 h after cell seeding, and a second transfection with theluciferase and miR expression plasmids was performed 24 h thereafter. At48 h after transfection with luciferase plasmids, cells were harvested byscraping and washed twice in ice-cold PBS. Cells from each culture dishwere suspended in 1 ml lysis buffer supplied by the manufacturer, flashfrozen in liquid nitrogen, and stored at �80°C. The cells were lysed using3 freeze-thaw cycles, and the lysates were clarified by centrifugation at13,500 � g for 5 min. The supernatants were assayed immediately accord-ing to the manufacturer’s instructions or stored at �20°C prior to assay.Luminescence was read using a microplate luminometer (Packard, Meri-den, CT). Three readings were taken for each sample, adjusted for back-ground luminescence, and averaged. The average luminescence valuesfrom three to six independent replicate cultures were also averaged toobtain each data point reported. A two-way Student’s t test was performedto determine whether observed differences in data were statistically sig-nificant (P � 0.05).

Affinity purification of MBPs and RNA binding assays. E. coli con-taining the pMal-5cE plasmid for expression of MBP or plasmid to ex-press an MBP fusion protein with the C-terminal domain (aa 215 to 340)of wild-type or mutated VP35 were grown to mid-log phase (A600 � 0.5)at 37°C. Protein expression was subsequently induced by the addition ofIPTG (isopropyl-�-D-thiogalactopyranoside) (0.3 mM). Cells from 1 literof culture were harvested 2 h after induction and suspended in 25 ml ofcolumn buffer (20 mM Tris-HCl, pH 7.5, 200 mM NaCl, 1 mM EDTA, 1mM sodium azide, and 10 mM 2-mercaptoethanol). The cells were lysedat 0°C by intermittent ultrasonic disruption in 10-s increments for ap-proximately 2 min of sonication time using a Branson model 350 sonica-tor microprobe at setting 5. After sonication, the cellular membranes wereremoved by centrifugation at 20,000 � g for 20 min at 4°C. The superna-tants (extracts) were collected and used immediately or were stored at�80°C prior to affinity purification of MBPs. Two-milliliter columns ofamylose resin (New England BioLabs, Beverly, MA) were prepared andequilibrated with column buffer according to the instructions of the man-ufacturer. Bacterial extracts (1 ml or equivalent to extract from 40-mlculture) were diluted 1:6 in column buffer and applied to amylose col-umns. Unbound proteins were removed by washing with 10 column vol-umes. 32P-labeled dsRNA or oligo(dT)20, prepared as described below,was added to columns containing the bound MBP or MBP fusion proteinand incubated at room temperature for 10 min. Unbound nucleic acidwas removed by washing with 10 volumes of column buffer. MBP andMBP fusion proteins, together with bound nucleic acid, were eluted by theaddition of 10 mM maltose in column buffer. A total of 30 fractions (0.4ml each) were collected, and 5 �l from each fraction was spotted onto aWhatman 3MM filter, dried, and counted by liquid scintillation spec-trometry. Nucleic acids were 5= end labeled with [�-32P]ATP and T4 bac-teriophage polynucleotide kinase and purified by standard procedures.RNA binding assays used 50 �g poly(I) · poly(C) for long dsRNA (specificactivity, 1.1 � 106 dpm/�g), 24 ng oligo(dT)20 (4.6 � 108 dpm/�g), or146 ng siRNA (2.6 � 108 dpm/�g) per column. The lengths of the poly(I) ·poly(C) duplexes (Sigma, St. Louis, MO) were variable. Oligo(dT)20 andsiRNA were obtained from IDT (Coralville, IA) and Ambion (Austin,TX), respectively. The siRNA contained two annealed 21-nucloetide (nt)RNAs with 19 perfect base pairs and 2-nt 3= single-strand overhangs ateach end.

Plant silencing suppression assays. Two- and three-component si-lencing suppressor assays have been described previously (23, 49). Briefly,wild-type Nicotiana benthamiana plants were used in the three-component assay, while the two-component assay utilized transgenic N.benthamiana plants expressing GFP from a Cauliflower mosaic virus 35Spromoter (line 16c; provided by David Baulcombe, Cambridge Univer-sity) (42). Agrobacterium harboring plasmids for expression of the GFPsilencing target, a long hairpin of GFP (dsGFP) to induce silencing, and/ora protein of interest or control were mixed at equal volumes (A600 � 1)and coinfiltrated into the underside of 3 leaves per plant using a 1-mlsyringe. In all cases, expression was driven by the 35S promoter. GFPfluorescence was analyzed 3, 5, and 7 days postinfiltration with a long-wave UV lamp (Blak-Ray Model B 100YP; UV Products). Photographs ofindividual leaves were obtained using a Nikon D40 camera with UV andyellow filters at a uniform exposure.

RNA extraction and Northern blot analysis. Northern blot proce-dures have been described previously (49). Leaf discs containing the infil-trated zones were obtained from 3 plants (weighing approximately 0.15 gtotal), and RNA was extracted using 1 ml TRIzol Reagent (Invitrogen).RNA loaded onto gels for Northern blots contained 5 �g for each sample,with the exception of p19 samples, for which 1 �g of RNA was loaded.Following RNA transfer, the membranes were probed with a GFP anti-sense riboprobe synthesized with [�-32P]UTP using the Maxiscript T7 kit(Ambion) according to the instructions of the manufacturer.

RESULTSEBOV VP35 can suppress microRNA-directed silencing inmammalian cells. Haasnoot and coworkers (15) were the first todescribe EBOV VP35 as an RSS. In their studies, short hairpinswere targeted to the open reading frame to silence firefly luciferasereporter activity, and EBOV VP35 was shown to suppress thissilencing. Inasmuch as gene silencing in mammalian cells occursby the targeting of cellular microRNA to predominantly untrans-lated sequences, and since microRNA silencing could alter thecellular environment to enhance virus replication, we wished toexamine the ability and specificity of EBOV VP35 to interfere withmicroRNA-directed silencing in mammalian cells. A modificationof the reporter assay developed by Zeng et al. was employed (57).The reporter construct consisted of the Fluc gene linked to 8 tan-dem repeats of the miR30 target sequence within the 3= UTR se-quences. When cotransfected with plasmid constructs encodingthe target Fluc and the irrelevant Rluc used as a transfection con-trol, the miR30-encoding plasmid severely diminished relativeFluc (Fluc/Rluc) activity compared to the activity observed whenno miR-expressing plasmid was added or when cells were cotrans-fected with an irrelevant miR21 in 2 different cell types (BHK and293T) (Fig. 1A and B, respectively).

Because our goal was to measure the ability of a viral protein tosuppress silencing activity, it was important to determine thedose-response of the silencing activity of the miR30 expressionconstruct. In preliminary experiments, we observed nearly com-plete silencing of Fluc activity when as little as 0.1 ng of miR30-encoding plasmid was used in some experiments, but the absoluteresponses were highly variable from experiment to experiment.We observed more consistent dose-responses when cells weretransfected with a constant amount of microRNA-expressingplasmid in which the ratio of specific (miR30-expressing) to non-specific (miR21-expressing) plasmid was varied. Examples ofdose-response curves in BHK and 293T cells are shown in Fig. 1Cand D, respectively. Regardless of the total amount (10 or 30 ng) ofmicroRNA-expressing plasmid or the type of cell line used, weobserved a negative logarithmic relationship between the relative

Zhu et al.

3040 jvi.asm.org Journal of Virology

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

dose of miR30-expressing plasmid and Fluc activity. In contrast,the level of activity of the irrelevant target gene product, Rluc,displayed no pattern of significant inhibition by miR30 (Fig. 1Cand D).

EBOV VP35 prevents the establishment of miR-directedRNA silencing in mammalian cells. We first used cotransfectionexperiments to determine the ability of EBOV VP35 to suppressmiR30-directed silencing of Fluc activity. For BHK cells, a mixture

FIG 1 Specificity and dose-response of microRNA-directed silencing in mammalian cells. Each dish of BHK (A and C) or human 293T (B and D) cells wascotransfected with 30 ng of the Fluc reporter plasmid (pCMV-Luc-miR30) containing miR30 target sequences in the 3= UTR and 2 ng of the Rluc control reporter(pCMV-Rluc) alone or together with plasmid expressing miR30 and/or miR21 as indicated. Fluc and Rluc activities were determined from cells harvested 48 hafter transfection using the dual-luciferase reporter kit as described in Materials and Methods. (A and B) Cells were incubated with reporter plasmids only (nomiR) or with 10 ng (A) or 30 ng (B) miR-expressing plasmid as indicated. The average Fluc/Rluc activities from 3 replicate samples were determined (� standarddeviation) and normalized to that obtained with no microRNA (set at 100%). (C and D) A dose-response for selective reactivity to miRs was determined bycotransfecting cells with reporter plasmids together with a constant amount of miR-expressing plasmids, varying the amount of specific miR30 (indicated on theabscissa) to the amount of the nonspecific miR21-expressing plasmid. (C and D) For BHK cells (C), total miR-expressing plasmid was 10 ng, and for 293T cells(D), total miR-expressing plasmid was 30 ng. The average Fluc (Œ) and Rluc (�) activities (� standard deviations) of triplicate samples are shown. (E) 293T cellswere transfected with protein-expressing plasmids as indicated (2 �g/60-mm dish) and 24 h later supertransfected with luciferase reporter plasmids and miR30silencing mixture (12 ng miR30- plus 78 ng miR21-expressing plasmids). Luciferase activities from six replicate samples receiving the silencing mixture wereaveraged and are shown as a percentage of that in samples that received miR21-expressing plasmid only (no silencing). The error bars indicate standarddeviations. The statistical significance of differences between samples containing the silencing miR30-expressing plasmid and those containing only the miR21-expressing plasmid (*, P � 0.05) was determined by a two-sided Student’s t test. (F) Immunoblots showing expression of test proteins, EGFP (�-EGFP), p19(�-Flag), or VP35 (�-myc) in 4 �g of extract from parallel cultures of silenced cells (lanes 2, 4, and 6). The same amount of control 293T cell extract was loadedonto the gels (lanes 1, 3, and 5) and probed with the indicated antibodies.



Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

containing 0.15 ng miR30 and 9.85 ng miR21 was selected. Thismixture incompletely silenced Fluc activity but left silencing ofFluc in a responsive range. The silencing mixture was cotrans-fected with expression plasmid encoding either EBOV VP35, anirrelevant protein (EGFP), or the plant virus RSS p19 linked to aFLAG epitope. Luciferase activity was compared to that in nonsi-lenced cells transfected with 10 ng miR21-expressing plasmid (seeFig. S1 posted at http://go.osu.edu/JVI5741-11Supplementary).Coexpression of EGFP and the silencing miR mixture resulted inan �2.5-fold reduction in Fluc activity compared to the nonsi-lenced control. However, coexpression of EBOV VP35 and thesilencing mixture restored Fluc activity. We also observed partialbut significant restoration of Fluc activity by the p19 plant virusRSS. Because all plasmids were expressed at the same time, theseresults could not differentiate between the ability of VP35 to pre-vent the establishment of or to reverse miR-directed silencing.

A silencing mixture of 4 ng miR30 and 26 ng miR21 for 35-mmplates (or 12 ng miR30 and 78 ng miR21 for 60-mm plates) wasselected as the most responsive ratio for silencing suppressionexperiments in 293T cells or their derivatives (see Fig. 1D). More-over, due to the high transfection efficiency of 293T cells, it waspossible to perform sequential transfections to determine whetherprior expression of VP35 or other proteins could prevent silencingby miR30. Cells were transfected with test expression plasmid 24 hprior to the addition of the reporter and miR expression plasmidsfor these experiments (Fig. 1E). The miR30-miR21 mixture re-duced Fluc expression with prior expression of EGFP compared tothat observed when only miR21 was expressed. However, not onlydid prior expression of VP35 protein prevent the silencing of 293Tcells, but VP35 expression led to even greater activity than wasobserved in nonsilenced cells. In contrast, we observed partialrecovery of Fluc expression when cells were preincubated with the

p19 plant virus RSS, though Fluc expression remained signifi-cantly below that observed in nonsilenced controls. Although thislow level of silencing suppression by p19 reached significance inthis experiment with 6 replicates, in others using only 3 replicates,we could discern no significant difference between Fluc activitywith prior expression of p19 and that with the EGFP control (notshown). Analysis of test protein expression in silenced samplesconfirmed that all proteins were well expressed (Fig. 1F). Thus, thelow-level silencing suppression by p19 observed in either 293T orBHK cells (Fig. 1E; see Fig. S1 posted at http://go.osu.edu/JVI5741-11Supplementary) is dwarfed by the high RSS activity ofVP35 and VP35’s ability to prevent the establishment of miR-directed silencing.

Ability of mutant EBOV VP35 to prevent miR-directed si-lencing in 293T cells. Because of the potent ability of miR30 tosilence Fluc expression, it was important to ensure that all cellsinto which miR30 plasmid was introduced also expressed the testprotein in order to accurately determine the abilities of mutatedversions of VP35 to suppress miR-directed silencing. Thus, weestablished cell lines (derived from 293T cells) that would induc-ibly express wild-type VP35 or a mutated form thereof. The E. colilacZ or EBOV VP35 gene was placed under the control of the TEToperator and stably introduced into 293T cells that constitutivelyexpress the TET repressor, as described in Materials and Methods.Expression of TET operator-controlled genes is tightly regulatedin these cell lines, since little or no protein was detected in theabsence of the inducer, DOX. However, following 24 h in thepresence of DOX (1 �g/ml), substantial expression of the con-trolled genes was observed (Fig. 2B and C). Visualization of X-Galstaining in lacZ-expressing cell lines demonstrated that gene ex-pression was induced in �95% of the cells under these conditions(results not shown).

FIG 2 Ability of induced wild-type or mutant VP35 to suppress miR-directed silencing in stable cell lines. (A) Protein expression in the stably transfected293T-derived cell lines was induced (open bars) by the addition of DOX (final concentration, 1 �g/ml) in culture medium or uninduced (cross-hatched bars) atthe same time by the addition of the equivalent amount of vehicle-containing medium (5% dimethyl sulfoxide [DMSO] in DMEM). Twenty-four hours later,the cells were transfected with luciferase reporter plasmid and either 30 ng/dish of control miR21 (not silenced) or a mixture of 4 ng miR30 plus 26 ng miR21(silenced). Fluc activity was measured in extracts of cells harvested 48 h later. The similarly derived cell line that inducibly expresses the irrelevant �-Gal protein(LacZ) was used as a control for the level of silencing that could be established in this system. A representative experiment showing the average amounts ofluciferase activity (normalized for the protein amount) from three replicate samples is shown. The fold enhancement of luciferase activity in DOX-inducedcompared to uninduced samples is indicated for wild-type (WT)- and VP35 mutant-expressing cell lines. The significance of the differences observed wasdetermined by a two-sided Student’s t test (P � 0.05) and is discussed in detail in the text. No significant difference in Fluc expression between induced anduninduced lacZ cells was observed. The error bars indicate standard deviations. (B and C) Protein expression in cells treated with the miR30 silencing complexis shown for uninduced (�) or DOX-induced (�) cells. Equivalent amounts of protein (4 �g) were separated by SDS-PAGE and immunoblotted with antibodyto �-Gal (B) or antibody to c-myc (C) to detect protein expression in LacZ or VP35 (wild-type or mutant)-expressing cells, respectively. M, molecular weightmarker that contains �-Gal; C, control VP35 expression in 293T cells transfected with VP35 expression plasmid.

Zhu et al.


Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

The Fluc target reporter plasmid was introduced into the lacZcell line, together with either 30 ng of miR21-expressing plasmid(not silenced) or a mixture of 4 ng miR30- and 26 ng miR21-expressing plasmid (silenced). The miR30-miR21 mixture was ca-pable of silencing Fluc expression more than 2-fold, regardless ofwhether lacZ expression was induced by the addition of DOX (Fig.2A). These results indicate the specificity of silencing by miR30and the absence of interference by inducible protein expression inthe establishment of silencing by miR30. The abilities of wild-typeand mutant VP35 genes to suppress the establishment of silencingby miR30 were tested in the relevant uninduced cell lines or thosein which gene expression was induced with DOX for 24 h prior totransfection. We observed enhanced Fluc activity in the absence ofinduction for some of the cell lines (K309A and R312A), suggest-ing the presence of small but functional amounts of VP35 proteindespite its lack of detection by immunoblotting (Fig. 2C). Thedifferent absolute levels of Fluc activity without DOX inductiontherefore likely reflect different basal levels of expression of theproteins. Nevertheless, we observed significantly increased Flucactivity in cells induced to express either the wild-type (3.8-fold)or K309A or R312A mutant (2.2- and 1.7-fold, respectively) VP35proteins compared to that expressed without induction (Fig. 2A).Moreover, the induced expression of wild-type VP35 or either ofthe VP35 mutants (K309A and R312A) significantly increasedFluc activity compared to that present in the induced lacZ controlcells, demonstrating that much, though not all, of the RSS activityin VP35 (wild-type or mutant)-expressing cell lines was depen-dent upon induced protein expression. Notably, the R312A andK309A mutations, reported to completely abolish the dsRNAbinding ability of VP35 (7, 12), possessed significant RSS activitycompared to the lacZ control. However, the RSS activity of theR312A mutant was significantly less than that observed followingthe induction of the wild-type or K309A mutant gene. Surpris-ingly, the Fluc activity in K309A-induced cells was as high as orhigher than that observed in cells induced to express the wild-typeVP35 (Fig. 2A). These differences were not due to differences inthe levels of induced wild-type or mutant VP35 protein expres-sion, since we observed comparable levels of protein expression inthe silenced samples upon DOX induction (Fig. 2C). Taken to-gether, these results indicate that at least some of the RSS activityof VP35 in mammalian cells is RNA binding independent. How-ever, because the RSS activities of the R312A and K309A mutantsdiffered and previous experiments did not investigate the abilitiesof these mutants to bind to small RNA species, it was important toclarify the abilities of mutant proteins to bind to small RNAsstructurally similar to those that mediate microRNA-directed si-lencing.

Ability of EBOV VP35 to bind siRNA. Although VP35 hasbeen shown to bind to blunt-ended 5=-phosphorylated and non-phosphorylated RNA (7, 19, 26, 30), its ability to bind to media-tors of silencing suppression (with 3= OH overhangs) has not beendetermined. siRNAs are capable of knocking down gene expres-sion in both plant and mammalian cells (reviewed in references34, 43, and 48) and, like microRNAs, are incorporated into RNA-induced silencing complexes (RISCs) to mediate posttranscrip-tional silencing of mRNA. Both siRNAs and microRNAs contain a19-nt duplex or partially duplex region with 3= overhangs of 2 nton each end. Since the VP35 CTD has been shown to be involvedin the dsRNA binding activities of the protein (7, 19, 30), we ex-pressed wild-type and mutant forms of this domain (amino acids

215 to 340) as an MBP fusion protein in bacteria. The functional-ity of purified MBP-VP35 CTD preparations (with a myc tag Nterminal to the CTD) was first confirmed. Affinity columns con-taining either MBP or the MBP-VP35 CTD linked to amylose wereprepared as described in Materials and Methods. Different formsof radioactively labeled nucleic acid were incubated with thebound proteins. MBP-containing proteins, together with nucleicacid bound to them, were then eluted by the addition of maltose.In control experiments, we demonstrated that this procedureeluted MBP or MBP-VP35 with a purity of �95% (see Fig. S2posted at http://go.osu.edu/JVI5741-11Supplementary). Asshown in Fig. 3A and D, neither MBP nor the MBP-VP35 CTDprotein was capable of binding a single-stranded DNA,oligo(dT)20. In contrast, MBP-VP35 CTD specifically boundpoly(I) · poly(C) (Fig. 3E), whereas MBP failed to bind this long(up to 2-kb) dsRNA (Fig. 3B). MBP-VP35 CTD, but not MBPalone (Fig. 3F and C, respectively), was also capable of bindingsiRNA. These results demonstrate that MBP-VP35 CTD bindsboth siRNA and larger dsRNA with high specificity under theseconditions. We also tested the abilities of various VP35 mutationswithin this context to bind siRNA. As expected, we observed nosignificant binding of the siRNA to the MBP-K309A CTD or theMBP-R312A CTD or to a double mutant containing the R312Aand K309A mutations (Fig. 3G to I). These results confirm thatwild-type VP35, or at least its CTD, is capable of stably bindingsiRNA with 2-nt 3= overhangs and that point mutations at R312and/or K309 eliminate this binding. Together with the observa-tion that both of these mutant proteins can enhance Fluc expres-sion following miR-directed silencing in mammalian cells, thoughto different extents, these results also suggest that most of thisability is independent of RNA binding activity.

EBOV VP35 functions as a silencing suppressor in plants.EBOV VP35 has been shown to be a potent IFN antagonist, capa-ble of blocking not only the induction of IFN-�, but also down-stream effectors of IFN, such as PKR (1). Indeed, translationalinhibition caused by the ability of activated PKR to phosphorylatethe essential translation factor eIF-2� is also antagonized by VP35(12, 45). Inasmuch as miR-directed silencing in mammalian cellscan lead to translational inhibition (35, 57), assessment of the RSSactivity of VP35 in mammalian cells using a reporter is compli-cated by the presence of the IFN pathway. In order to evaluate theinherent RSS activity of VP35 (wild type and mutants) indepen-dent of any effects it has on the IFN pathway, we used two well-established plant silencing suppressor assays (23), since plants nei-ther encode IFNs nor respond to them. In the three-componentsystem, potent silencing was established by introduction (viaAgrobacterium infiltration of N. benthamiana leaves) of plasmidsexpressing the GFP gene and a long hairpin of GFP (dsGFP), andsilencing suppression was determined by the simultaneous addi-tion of a test gene. Expression of the GFP gene is silenced when thedsGFP is converted to short siRNAs, which then associate withRNA-silencing complex proteins (16, 23, 55). The siRNAs mayalso be amplified by a plant-encoded RNA-dependent RNA poly-merase that strengthens and facilitates the spread of the silencingresponse. In the two-component system, a GFP-transgenic line ofN. benthamiana (line 16c) was used, and a GFP expression clone(which induces silencing of GFP mRNA originating from thetransgene and the expression plasmid) was introduced at the sametime as a test gene for silencing suppression. When plant leaves areexamined under UV light, they appear red due to the inherent



Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

fluorescence of chlorophyll, and expression of GFP in the areas ofinfiltration is evident by the bright green color. In both systems,silencing developed over a period of several days, as indicated bythe near absence of green color when the leaves were coinfiltratedwith the negative-control test plasmid encoding GUS, whichfailed to suppress silencing (Fig. 4A and B). In contrast, GFP ex-pression persisted even after 7 days when leaves were coinfiltratedwith the potent plant virus RSS p19. The wild-type VP35 gene,expressed using the same strong 35S promoter, also suppressed

silencing. As an additional control, we tested a G333S VP35 mu-tant, which was derived by the same PCR-directed mutagenesisand domain switching used to derive all of the mutant expressionclones required for these experiments; however, the G333 residuelies outside the central patch of basic residues associated with RNAbinding (30). As expected, G333S displayed RSS activity similar tothat displayed by wild-type VP35 (Fig. 4A and B). Compared tothe negative GUS control, low levels of GFP expression were stillobserved within the areas of infiltration in the presence of K309A

FIG 3 RNA binding by the C-terminal domain of wild-type and mutant VP35 proteins. MBP (A to C) or MBP linked to the C-terminal domain (aa 215 to 340)of wild-type EBOV VP35 (MBP-VP35CTD) (D to F) or mutants containing K309A (G), R312A (H), or both (K/R) (I) mutations were expressed in E. coli.Clarified extracts of cells were bound to a 2-ml amylose column, and unbound proteins were removed by extensive washing as described in Materials andMethods. Radioactive nucleic acid, as indicated, was added to the column, followed by washes, and radioactivity bound to MBP or MBP fusion protein was elutedby the addition of 10 mM maltose in binding buffer. Fractions (0.4 ml) were collected, and a 5-�l sample of each fraction was counted by liquid scintillationspectrometry. (A and D) 20-mer oligo(dT) (dT20). (B and E) Poly(I) · poly(C) (dsRNA). (C, F, G, H, and I) siRNA.

Zhu et al.


Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

and R312A mutant VP35 genes, particularly in the two-component system (Fig. 4B). However, both of these mutationsclearly compromised the RSS activity of VP35. The double mutantreduced the RSS activity even further in the two-component sys-

tem, though low residual GFP expression remained observableeven after 7 days (Fig. 4B).

The visual data in these assays were generally confirmed byNorthern blots to determine changes in the steady-state levels of

FIG 4 EBOV VP35 suppression of silencing in plants. (A) In this three-component assay, N. benthamiana plant leaves were infiltrated with equivalent amountsof Agrobacterium containing plasmids encoding GFP as a silencing target, a long dsGFP hairpin as the silencing inducer, and a test protein for silencingsuppression as indicated. Expression of target, hairpin, and test genes were all under the control of the Cauliflower mosaic virus 35S constitutive promoter. Ondays 3, 5, and 7 postinfiltration, leaves were examined under UV light, and all leaves harvested on a given day were photographed at the same exposure level.Under UV light, the chlorophyll appears red, whereas GFP expression is green. The negative-control test protein for silencing suppression was GUS, and thepositive-control silencing suppressor was p19 encoded by Cymbidium ringspot virus. Both WT and mutant forms of VP35 were individually tested. The resultsshown are representative leaves from at least 3 independent experiments in which 3 leaves of 10 plants were inoculated for each condition. (B) In thistwo-component assay, a GFP-transgenic line (line 16c) of N. benthamiana primed for silencing of GFP was used. Leaves were infiltrated with agrobacteriacontaining a plasmid that expresses GFP to induce silencing and a test plasmid for silencing suppression. Only leaves harvested on day 7 are shown. (C and D)Northern blots of RNA extracted from infiltrated leaf material from a three-component (C) or two-component (D) assay system. RNA was denatured andseparated by electrophoresis through 1.2% agarose gels containing 1.3% formaldehyde. The RNA was transferred to a membrane and probed with [32P]GFPsingle-stranded RNA (ssRNA) antisense probe as indicated in Materials and Methods. For all samples except those containing the p19 plasmid, equivalentamounts of total RNA were loaded, as exhibited by ethidium bromide staining of 28S rRNA shown in the lower blots. Due to the high level of silencingsuppression by p19 (and the corresponding high level of expression of GFP mRNA), the gels were loaded with 20% of the amount of RNA in other samples toprevent overexposure of the blots.



Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

GFP mRNA within the areas of infiltration. In plants, siRNA gen-erated in response to silencing results in specific cleavage of thetarget mRNA, leading to reduced steady-state levels. We isolatedRNA from infiltrated spots 7 days after inoculation, separated theRNA by denaturing agarose gel electrophoresis, and probed trans-ferred RNA with a GFP riboprobe. Northern blots of the RNAfrom infiltrated tissues from the three- and two-component assaysare shown in Fig. 4C and D, respectively. The RNA load is indi-cated by the ethidium bromide staining of 28S rRNA in the gelprior to transfer. Because of p19’s potent RSS activity (46), 5-foldless RNA was loaded for p19-containing samples to prevent over-exposure of the blots. As expected, an extremely low steady-statelevel of GFP mRNA (barely detectable in the three-componentsystem) was present in plants that received the negative control(GUS) plasmid compared to the levels in plants that received thepositive-control p19 RSS or wild-type VP35 (Fig. 4C and D). Ofthe VP35 mutants tested, only G333S possessed steady-state levelsof GFP mRNA that were significantly greater than those ofnegative-control samples, confirming that in the presence of wild-type or G333S VP35, GFP mRNA degradation associated withsilencing was inhibited. These results are consistent with the visualinspection of GFP expression in the plants (Fig. 4A and B) anddemonstrate the ability of EBOV VP35 to act as an RSS in plantsindependent of the known effects of the protein on IFN response.Moreover, the results show that mutation of the R312 and/orK309 residue, which largely destroys the ability of VP35 to bind tosiRNA, also eliminates most of the plant RSS activity. Thus, ex-pression of EBOV VP35 in plants can either prevent the establish-ment of silencing and/or rapidly reverse it, and this response ishighly dependent on the ability of the protein to bind siRNA orother intermediate RNA species in the silencing pathway.

DISCUSSIONSuppression of miR-directed silencing by EBOV VP35 in mam-malian cells. In this report, we analyzed in greater detail the si-lencing suppressor activity of wild-type and mutant EBOV VP35first reported by Haasnoot and colleagues (15). Our results dem-onstrate that wild-type VP35 can specifically suppress microRNA-directed silencing in mammalian cells in which the target se-quence is located in the 3= UTR of the reporter gene, the site wheremost microRNA targets are found. Our results also demonstratethat some mammalian cell lines differ in their relative susceptibil-ities to miR-directed silencing. In our hands, BHK cells are exqui-sitely sensitive to silencing, with �1 ng miR30-expressing plasmidresulting in a 5-fold reduction in target protein expression (Fig.1C). Although miR-directed silencing of human 293T cells alsooccurs, similar reductions in reporter silencing required transfec-tion with much larger amounts of miR30-expressing plasmid(compare Fig. 1A and B). By mixing the miR30- and miR21-expressing silencing plasmids, specificity of silencing response andbetter reproducibility were ensured (Fig. 1C and D). Moreover,dose-response curves suggested ranges of silencing plasmids capa-ble of distinguishing relative RSS abilities of test proteins. By usinga small dose of miR30 relative to miR21 plasmid, VP35 consis-tently suppressed the silencing of target gene expression in both293T and BHK cells (Fig. 1E; see Fig. S1 posted at http://go.osu.edu/JVI5741-11Supplementary). The potent plant RSSp19, known to bind siRNA (6, 46), possessed low, but significant,RSS activity in BHK and 293T cells, though only partial recoveryfrom silencing occurred in both cell types. The p19 protein was

shown previously to possess some RSS activity in miR-directedsilencing assays in yet another cell type, HeLa cells (39). Despitesome quantitative differences in the amounts of silencing suppres-sion observed in BHK and 293T cells, our results demonstrateconclusively that VP35 can prevent the establishment of miR-directed silencing.

To determine the relative differences in the activities of wild-type versus mutant VP35 genes, it was important to control forcell-line- or species-specific differences, as well as to ensure thatessentially all of the cells receiving silencing plasmids also ex-pressed the test RSS. This was achieved by deriving stable induc-ible cell lines from a single clone of 293T cells that constitutivelyexpresses the TET repressor. These stable cell clones displayedlittle if any protein expression (detectable by immunoblotting) inthe absence of DOX induction (Fig. 2B and C). Introduction of themiR30/miR21-expressing plasmid mixture established effectivesilencing regardless of the addition of DOX, as shown in the lacZ-expressing clones (Fig. 2A). To reduce the effects of clonal differ-ences, we compared the responses of specific clones with or with-out induction of test protein expression prior to the establishmentof silencing. Based on that comparison, we found that the wild-type VP35 and both mutant VP35 proteins permitted significantlymore reporter protein expression following DOX induction thanwas observed in noninduced cells. Previous analysis of VP35 func-tion demonstrated that the K309A and R312A mutant formscould abrogate PKR activity and eIF-2� phosphorylation as effi-ciently as the wild-type protein, resulting in enhanced (nonspe-cific) protein expression (12, 45). Thus, the ability of the R312Amutant protein to increase Fluc expression at significantly lowerlevels than either the wild-type or K309A mutant in our reporterassays suggests that it is the RSS activity of R312A that is compro-mised in the miR-directed assay. In other words, the enhancementof reporter activity by R312A is likely due to the nonspecific trans-lational effect of VP35 caused by its PKR antagonism, whereas thecumulative effect of VP35 in mammalian systems (by the wild typeand the K309A mutant) reflects both specific RSS activity andnonspecific relief of translational inhibition caused by stress-induced kinases. This interpretation would explain the higher lev-els of reporter expression observed in silenced cells that expressVP35 (wild type or K309A) than in cells that were not silenced(miR21 only) or were silenced but expressed an irrelevant gene(Fig. 1E and 2A).

Our ability to observe wild-type levels of microRNA-directedsilencing suppression by the K309A mutant differs from the re-sults of Haasnoot and coworkers, who concluded that neitherK309A nor R312A could suppress silencing by short hairpin tran-scripts (15). Since these investigators introduced the VP35-expressing plasmids at the same time as the reporter and silencingconstructs, silencing could have been established prior to signifi-cant expression of VP35 protein, or expression of VP35 proteincould have been significantly less than we obtained in our studies.In our hands, we observed the most potent silencing suppressionwhen VP35 proteins were present prior to the introduction ofsilencing plasmids, either by transfecting the VP35 expressionplasmid a day prior to transfection with silencing mixtures (Fig.1E) or by expressing VP35 proteins in stable cell lines by the ad-dition of DOX for 24 h prior to the establishment of silencing (Fig.2A). Moreover, our use of inducible cell lines resulted in VP35protein (wild type or mutant) expression in �95% of the cells thatwere transfected with the silencing mixture. Thus, the inability of

Zhu et al.


Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

Haasnoot et al. to observe silencing suppression by K309A (15)may be due to the failure of the mutant to reverse establishedsilencing, failure to express the mutant protein at sufficientlyhigh levels in most cells, and/or failure to overcome what couldbe more potent silencing than was present in our experiments.

It is important to note that the difference between themicroRNA-directed silencing suppression activities of K309A andR312A mutants cannot be explained by differences in protein ex-pression levels (Fig. 2C) or by their abilities to bind small siRNAs,since neither mutant CTD could bind siRNAs (Fig. 3G and H).Thus, our results suggest that the microRNA-specific silencingsuppression activity of VP35 observed in mammalian cells is notdependent on its ability to bind small RNAs. These results areconsistent with a recent report that demonstrated the ability of theRISC-associated proteins PACT and TRBP to bind to EBOV VP35when overexpressed by transfection (11). These investigators alsoreported that the VP35 interactions with PACT and TRBP werenot dependent upon the presence of siRNA. Thus, it is likely thatVP35 can sequester important RISC components, thereby pre-venting the assembly or activity of silencing complexes.

PACT has also been shown to activate PKR during stress, lead-ing to an inhibition of translation (32). The fact that VP35 canassociate with and presumably block the activity of PACT mayexplain the translational enhancement caused by VP35 that is in-dependent of silencing suppression (for example, with the R312Amutant [Fig. 2A]) and that results in greater reporter expressionthan is observed in nonsilenced cells (Fig. 1E and 2A). It is note-worthy that both we (Fig. 2A) and Fabozzi and colleagues (11)have observed that the silencing-independent translational en-hancement caused by VP35 is much less than its effect on eithermicroRNA- or siRNA-directed silencing, confirming the utility ofthe Fluc expression assays to measure microRNA-directed silenc-ing suppression by VP35.

Suppression of RNA silencing in plants. We show for the firsttime that EBOV VP35 can suppress silencing in plants usingtwo well-characterized plant silencing suppression assays, fur-ther confirming that silencing suppression can occur indepen-dently of IFN pathway responses. This increases the number ofmammalian virus proteins, including influenza virus NS1 andHIV tat protein, that have been shown to have RSS activity inplants (5, 9, 39). However, the structure of the VP35 CTDdiffers markedly from that of other viral RSS proteins or ofcanonical RNA binding proteins (26, 29–31). The two silencingsuppression assays differ in the mechanisms by which silencingis induced (23). In addition, the strengths of the silencing sig-nals differ, which most likely accounts for the residual mildgreen color in the two-component assay with the K309A andR312A single mutants but not with the double mutant (Fig.4B). The two-component system relies on the use of GFP-transgenic plants that are “primed” for silencing by expressionof a target transgene, and silencing is triggered by overexpres-sion of the same mRNA from an expression plasmid. This si-lencing signal is likely to be weaker than that generated fromthe long GFP hairpin transcript that induces silencing in thethree-component system (23). Nevertheless, both the two- andthree-component assays rely on amplification of RNA species(transcripts, hairpins, and/or siRNAs) mediated by host RNA-dependent RNA polymerase (31).

In contrast to microRNA-directed silencing in mammaliansystems, the RSS activity of VP35 in plants (Fig. 4) appears to be

largely dependent upon VP35 dsRNA binding activity. Thus,only wild-type (but not K309A or R312A mutant) VP35 caneffectively bind the siRNA effectors (Fig. 3) and inhibit silenc-ing. Nevertheless, the milder two-component silencing systemdid permit a low level of GFP expression when the single pointmutants were introduced, even after 7 days. The low-level ac-tivity suggests that some portion of VP35 RSS activity in plantsis independent of its ability to bind short dsRNA molecules.However, since VP35 may interact poorly with the heterolo-gous proteins that make up RISCs in plants, VP35’s ability tosuppress silencing through protein-protein interactions inplants is not as great as in mammalian cells.

Do silencing suppression mechanisms in plants and mam-mals differ? Since plants have a potent mechanism to amplifysiRNAs and silencing in plants is largely dependent on RNAsignal amplification or assembly of functional RISCs (31), it isnot surprising that plant-silencing suppression by EBOV VP35and influenza virus NS1 corresponds to the ability to binddsRNAs (5, 7). This is also true of p19 and many other plantRSS proteins, although some also interact with silencing path-way proteins (6). In mammals, although silencing is dependentupon small dsRNAs (microRNAs or siRNAs), there is no RNA-dependent RNA polymerase to amplify either these effectormolecules or RNA inducers of the pathways. It is interestingthat all mammalian virus proteins reported to have RSS activ-ity, including VP35, influenza virus NS1, and vaccinia virusE3L, are also PKR or IFN antagonists (5, 12, 13, 31, 33, 45). Assuggested by Fabozzi and colleagues, the fact that VP35 inter-acts with PACT and TRBP, both of which are important effec-tors in PKR and silencing complexes, may explain this concor-dance (11). In animals, could protein signaling molecules, suchas IFN, have evolved to amplify the ability of the cells to silencegene expression in the absence of RNA amplification? We havefound that a mutated VP35 (R312A) that loses its ability to bindto siRNA (Fig. 3H) but retains PKR antagonism (45) loses mostof its ability to counter microRNA-dependent silencing inmammalian cells (Fig. 2) and most of its ability to suppressRNA silencing in plants (Fig. 4). Given the fact that the RNAbinding domains of many proteins are also protein bindingdomains (44), it is possible that the R312A mutation impactsboth protein and RNA binding activities of VP35. Indeed, thecrystal structure of VP35 with dsRNA shows that the residue isimportant for interaction between the two monomers arrangedin an asymmetric dimer, as well as for backbone RNA binding(26, 30). Since functional silencing complexes in both plantand animal cells require small RNAs as well as proteins thatbind to each other and to siRNA or microRNA in the complex,uncovering the precise mechanism(s) of silencing suppressionin each system will require the availability of mutants that dif-ferentially disrupt RNA binding but not binding to a particularprotein target.

ACKNOWLEDGMENTS

We thank all members of the Parris and Bisaro laboratories for manyhelpful discussions. In particular, we thank Carlee Schaefer and EricBauer for expert preparation of many of the plasmids used for thesestudies.

This work was supported in part by a developmental grant from theGreat Lakes Regional Center of Excellence for Biodefense and EmergingInfectious Diseases Research (D.S.P. and D.M.B.) and by NSF grant MCB-



Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

0743261 (D.M.B.). N.C.C. was supported by NIAID fellowship F32AI082884 during part of this work. J.N.J. was supported by fellowshipsfrom the OSU Center for RNA Biology and from Pelotonia.

REFERENCES1. Basler CF, Amarasinghe GK. 2009. Evasion of interferon responses by

Ebola and Marburg viruses. J. Interferon Cytokine Res. 29:511–520.2. Basler CF, et al. 2003. The Ebola virus VP35 protein inhibits activation of

interferon regulatory factor 3. J. Virol. 77:7945–7956.3. Basler CF, et al. 2000. The Ebola virus VP35 protein functions as a type I

IFN antagonist. Proc. Natl. Acad. Sci. U. S. A. 97:12289 –12294.4. Brodersen P, Voinnet O. 2006. The diversity of RNA silencing pathways

in plants. Trends Genet. 22:268 –280.5. Bucher E, Hemmes H, de Haan P, Goldbach R, Prins M. 2004. The

influenza A virus NS1 protein binds small interfering RNAs and sup-presses RNA silencing in plants. J. Gen. Virol. 85:983–991.

6. Burgyan J, Havelda Z. 2011. Viral suppressors of RNA silencing. TrendsPlant Sci. 16:265–272.

7. Cardenas WB, et al. 2006. Ebola virus VP35 protein binds double-stranded RNA and inhibits alpha/beta interferon production induced byRIG-I signaling. J. Virol. 80:5168 –5178.

8. Chang T-H, et al. 2009. Ebola Zaire virus blocks type I interferon pro-duction by exploiting the host SUMO modification machinery. PLoS Pat-hog. 5:e1000493.

9. Delgadillo MO, Saenz P, Salvador B, Gracia JA, Simon-Mateo C. 2004.Human influenza virus NS1 protein enhances viral pathogenicity and actsas an RNA silencing suppressor in plants. J. Gen. Virol. 85:993–999.

10. Diaz-Pendon JA, Ding S-W. 2008. Direct and indirect roles of viralsuppressors of RNA silencing in pathogenesis. Annu. Rev. Phytopathol.46:303–326.

11. Fabozzi G, Nabel CS, Dolan MA, Sullivan NJ. 2011. Ebolavirus proteinssuppress the effects of small interfering RNA by direct interaction with themammalian RNA interference pathway. J. Virol. 85:2512–2523.

12. Feng Z, Cerveny M, Yan Z, He B. 2007. The VP35 protein of Ebola virusinhibits the antiviral effect mediated by double-stranded RNA-dependentprotein kinase PKR. J. Virol. 81:182–192.

13. Gale M, Jr, Katze MG. 1998. Molecular mechanisms of interferon resis-tance mediated by viral-directed inhibition of PKR, the interferon-induced protein kinase. Pharmacol. Ther. 78:29 – 46.

14. Geisbert TW, et al. 2010. Postexposure protection of non-human pri-mates against a lethal Ebola virus challenge with RNA interference: aproof-of-concept study. Lancet 375:1896 –1905.

15. Haasnoot J, et al. 2007. The Ebola virus VP35 protein is a suppressor ofRNA silencing. PLoS Pathog. 3:e86.

16. Hamilton AJ, Baulcombe DC. 1999. A species of small antisense RNA inposttranscriptional gene silencing in plants. Science 286:950 –952.

17. Harcourt BH, Sanchez A, Offermann MK. 1999. Ebola virus selectivelyinhibits responses to interferons, but not to interleukin-1b, in endothelialcells. J. Virol. 73:3491–3496.

18. Hartman AL, et al. 2008. Inhibition of IRF-3 activation by VP35 is criticalfor the high level of virulence of Ebola virus. J. Virol. 82:2699 –2704.

19. Hartman AL, Towner JS, Nichol ST. 2004. A C-terminal basic amino acidmotif of Zaire Ebolavirus VP35 is essential for type I interferon antago-nism and displays high identity with the RNA-binding domain of anotherinterferon antagonist, the NS1 protein of influenza A virus. Virology 328:177–184.

20. Hensley LE, Jones SM, Feldmann H, Jahrling PB, Geisbert TW. 2005.Ebola and Marburg viruses: pathogenesis and development of counter-measures. Curr. Mol. Med. 5:761–772.

21. Huang Y, Xu L, Sun Y, Nabel G. 2002. The assembly of Ebola virusnucleocapsid requires virion-associated proteins 35 and 24 and posttrans-lational modification of nucleoprotein. Mol. Cell 10:307–316.

22. Jahrling PB, et al. 1999. Evaluation of immune globulin and recombinantinterferon-alpha2b for treatment of experimental Ebola virus infections. J.Infect. Dis. 179:S224 –S234.

23. Johansen LK, Carrington JC. 2001. Silencing on the spot: induction andsuppression of RNA silencing in the Agrobacterium-mediated transientexpression system. Plant Physiol. 126:930 –938.

24. Kash JC, et al. 2006. Global suppression of the host antiviral response byEbola- and Margviruses: increased antagonism of the type I interferonresponse is associated with enhanced virulence. J. Virol. 80:3009 –3020.

25. Kasschau KD, et al. 2003. P1/HC-Pro, a viral suppressor of RNA silenc-

ing, interferes with Arabidopsis development and microRNA function.Dev. Cell 4:205–217.

26. Kimberlin CR, et al. 2010. Ebolavirus VP35 uses a bimodal strategy tobind dsRNA for innate immune suppression. Proc. Natl. Acad. Sci. U. S. A.107:314 –319.

27. Kobinger GP, et al. 2006. Chimpanzee adenovirus vaccine protectsagainst Zaire Ebola virus. Virology 346:394 – 401.

28. Lakatos L, et al. 2006. Small RNA binding is a common strategy tosuppress RNA silencing by several viral suppressors. EMBO J. 25:2768 –2780.

29. Leung DW, et al. 2009. Structure of the Ebola VP35 interferon inhibitorydomain. Proc. Natl. Acad. Sci. U. S. A. 106:411– 416.

30. Leung DW, et al. 2010. Structural basis for dsRNA recognition andinterferon antagonism by Ebola VP35. Nat. Struct. Mol. Biol. 17:165–172.

31. Li F, Ding S-W. 2006. Virus counterdefense: diverse strategies for evadingthe RNA-silencing immunity. Annu. Rev. Microbiol. 60:503–531.

32. Li S, et al. 2006. Molecular basis for PKR activation by PACT or dsRNA.Proc. Natl. Acad. Sci. U. S. A. 103:10005–10010.

33. Li W-X, et al. 2004. Interferon antagonist proteins of influenza andvaccinia viruses are suppressors of RNA silencing. Proc. Natl. Acad. Sci.U. S. A. 101:1350 –1355.

34. Marques JT, Carthew RW. 2007. A call to arms: coevolution of animalviruses and host innate immune responses. Trends Genet. 23:359 –364.

35. McManus MT, Sharp PA. 2002. Gene silencing in mammals by smallinterfering RNAs. Nat. Rev. Genet. 3:737–747.

36. Muhlberger E, Lotfering B, Klenk H-D, Becker S. 1998. Three of the fournucleocapsid proteins of Marburg virus, NP, VP35, and L, are sufficient tomediate replication and transcription of Marburg virus-specific monocis-tronic minigenomes. J. Virol. 72:8756 – 8764.

37. Muhlberger E, Weik M, Volchkov VE, Klenk H-D, Becker S. 1999.Comparison of the transcription and replication strategies of Marburgvirus and Ebola virus by using artificial replication systems. J. Virol. 73:2333–2342.

38. Prins KC, Cardenas WB, Basler CF. 2009. Ebola virus protein VP35impairs the function of interferon regulatory factor-activating kinasesIKK� and TBK-1. J. Virol. 83:3069 –3077.

39. Qian S, et al. 2009. HIV-1 Tat RNA silencing suppressor activity is con-served across kingdoms and counteracts translational repression ofHIV-1. Proc. Natl. Acad. Sci. U. S. A. 106:605– 610.

40. Raja P, Wolf JN, Bisaro DM. 2010. RNA silencing directed againstgeminiviruses: post-transcriptional and epigenetic components. Biochim.Biophys. Acta 1799:337–351.

41. Richardson JS, et al. 2009. Enhanced protection against Ebola virus me-diated by an improved adenovirus-based vaccine. PLoS One 4:e5308.

42. Ruiz MT, Voinnet O, Baulcombe DC. 1998. Initiation and maintenanceof virus-induced gene silencing. Plant Cell 10:937–946.

43. Saksela K. 2003. Human viruses under attack by small inhibitory RNA.Trends Microbiol. 11:345–347.

44. Saumet A, Lecellier C-H. 2006. Anti-viral RNA silencing: do we look likeplants? Retrovirology 3:3–14.

45. Schumann M, Gantke T, Muhlberger E. 2009. Ebola virus VP35 antag-onizes PKR activity through its C-terminal interferon inhibitory domain.J. Virol. 83:8993– 8997.

46. Silhavy D, et al. 2002. A viral protein suppresses RNA silencing and bindssilencing-generated 21- to 25-nucleotide double-stranded RNAs. EMBOJ. 21:3070 –3080.

47. Sun Y, et al. 2009. Protection against lethal challenge by Ebola virus-likeparticles produced in insect cells. Virology 383:12–21.

48. Voinnet O. 2005. Induction and suppression of RNA silencing: insightsfrom viral infections. Nat. Rev. Genet. 6:206 –220.

49. Wang H, Buckley KJ, Yang X, Buchmann RC, Bisaro DM. 2005.Adenosine kinase inhibition and suppression of RNA silencing by gemi-nivirus AL2 and L2 proteins. J. Virol. 79:7410 –7418.

50. Warren TK, et al. 2010. Advanced antisense therapies for postexposureprotection against lethal filovirus infections. Nat. Med. 16:991–994.

51. Wu Z, Zhu Y, Bisaro DM, Parris DS. 2009. Herpes simplex virus type 1suppresses RNA-induced gene silencing in mammalian cells. J. Virol. 83:6652– 6663.

52. Yao F, Winkler STT, Lu M, Eriksson C, Eriksson E. 1998. Tetracyclinerepressor, tetR, rather than the tetR-mammalian cell transcription factorfusion derivatives, regulates inducible gene expression in mammaliancells. Hum. Gene Ther. 9:1939 –1950.

Zhu et al.


Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

53. Yao F, Theopold C, Hoeller D, Bleiziffer O, Lu Z. 2006. Highly efficientregulation of gene expression by tetracycline in a replication-defectiveherpes simplex viral vector. Mol. Ther. 13:1133–1141.

54. Yi R, Qin Y, Macara IG, Cullen BR. 2003. Exportin-5 mediates thenuclear export of pre-microRNAs and short hairpin RNAs. Genes Dev.17:3011–3016.

55. Zamore PD, Tuschl T, Sharp PA, Bartel DP. 2000. RNAi: dsRNA directs

the ATP-dependent cleavage of mRNA at 21 to 23 nucleotide intervals.Cell 101:25–33.

56. Zeng Y, Cullen BR. 2003. Sequence requirements for micro RNA pro-cessing and function in human cells. RNA 9:112–123.

57. Zeng Y, Yi R, Cullen BR. 2003. MicroRNAs and small interfering RNAscan inhibit mRNA expression by similar mechanisms. Proc. Natl. Acad.Sci. U. S. A. 100:9779 –9784.



Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

vi o

n 10

Dec

embe

r 20

21 b

y 12

3.21

3.19

4.21

7.

http://jvi.asm.org

characterization of the rna silencing suppression activity of the

Documents