PLK1 Down-Regulates Parainfluenza Virus 5 Gene Expression

PLK1 Down-Regulates Parainfluenza Virus 5 GeneExpressionDengyun Sun1, Priya Luthra1, Zhuo Li2, Biao He1,2,3*

1 Intercollege Graduate Program in Cell and Developmental Biology, Pennsylvania State University, University Park, Pennsylvania, United States of America, 2 Department

of Veterinary and Biomedical Sciences, Pennsylvania State University, University Park, Pennsylvania, United States of America, 3 Center of Molecular Immunology and

Infectious Disease, Pennsylvania State University, University Park, Pennsylvania, United States of America

Abstract

The paramyxoviruses are a family of negative-sense RNA viruses that includes many important human and animal pathogens.Paramyxovirus RNA synthesis requires the viral phosphoprotein (P) and the large (L) protein. Phosphorylation of P is thought toregulate viral gene expression, though direct proof remains elusive. Recently, we reported that phosphorylation of a specificresidue (Ser157) of the P protein of parainfluenza virus 5 (PIV5), a prototypical paramyxovirus, correlates with decreased viralgene expression and cytokine expression in infected cells. Here, we show that: Polo-like kinase 1 (PLK1), a serine/theroninekinase that plays a critical role in regulating the cell cycle, interacts with PIV5 P through the S157 residue; PLK1 inhibitionincreases viral gene expression; PLK1 over-expression inhibits viral gene expression; and PLK1 directly phosphorylates P invitro, indicating that PLK1 down-regulates viral gene expression by phosphorylating P. Furthermore, we have determined thePLK1 phosphorylation site on P and found that mutant recombinant PIV5 whose P proteins cannot either bind to or bephosphorylated by PLK1 have similar phenotypes. Increased viral gene expression in PIV5 with mutations in the PLK1 binding/phosphorylation sites correlates with increased induction of cell death and cytokine expression, suggesting that PIV5 limits itsviral gene expression to avoid these host effects. It is possible that targeting PLK1 will enhance host innate immune responses,leading to a novel strategy of clearing paramyxovirus infections quickly.

Citation: Sun D, Luthra P, Li Z, He B (2009) PLK1 Down-Regulates Parainfluenza Virus 5 Gene Expression. PLoS Pathog 5(7): e1000525. doi:10.1371/journal.ppat.1000525

Editor: Christopher F. Basler, Mount Sinai School of Medicine, United States of America

Received January 12, 2009; Accepted June 30, 2009; Published July 24, 2009

Copyright: � 2009 Sun et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricteduse, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This work was supported by grants from the National Institute of Allergy and Infectious Diseases to B.H. (R01 AI070847 and K02 AI65795). The fundershad no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing Interests: The authors have declared that no competing interests exist.

* E-mail: [email protected]

Introduction

Viruses in the Paramyxoviridae family of Mononegavirales include

many important human and animal pathogens such as the human

parainfluenza viruses, Sendai virus (SeV), mumps virus (MuV),

Newcastle disease virus (NDV), measles virus (MeV), rinderpest

virus and human respiratory syncytial virus (RSV) as well as the

emerging viruses Nipah and Hendra virus. The paramyxovirus

RNA-dependent RNA polymerase (RdRp), which both transcribes

and replicates the viral RNA genome, consists of two proteins, the

phosphoprotein (P) and the large (L) protein [1]. While

paramyxovirus P proteins are all heavily phosphorylated (hence

the name phosphoprotein) and are essential for viral gene

expression, the role of P phosphorylation in the replication of

paramyxoviruses remains an enigma. Conclusive evidence on the

role of phosphorylation of the P protein in replication of

paramyxoviruses remains elusive. The most recent work seems

to indicate that the phosphorylation of the P proteins of

paramyxoviruses does not have a role in viral gene expression.

The best-studied P proteins of paramyxoviruses are the P proteins

of RSV and SeV. It was first reported in the 1970s that the P

protein of SeV is phosphorylated [2]. While as many as 11

phosphorylation sites were detected, the serine (Ser) residue at

position 249 was determined to be the major phosphorylation site

[3]. However, recombinant SeV containing mutations at the

major P phosphorylation sites have similar growth characteristics

and pathogenicity in vitro (cultured cells) and in vivo (mice) [4],

indicating that these sites are not important for viral gene

expression. Mutating five additional phosphorylation sites besides

S249 results in a P mutant whose level of phosphorylation is

reduced by more than 90% in transfected cells; yet, the mutant P

still has normal activity in a mini-genome system [5]. The P

protein of RSV is the most heavily phosphorylated of the

paramyxovirus P proteins [6]. Two clusters of phosphorylation

sites (amino acid residues 116, 117 and 119 and residues 232 and

237) have been identified [7–10]. When mutations are introduced

into these sites in recombinant RSV by a reverse genetics system,

expression levels of the viral genes are not adversely affected,

indicating that these residues do not play a critical role in viral

gene expression [11]. Further studies of the P protein using mass

spectrometry identified the threonine residue at position 108 as

being phosphorylated. The phosphorylation of T108 is important

for its interaction with M2-1, a processivity factor of viral RNA

synthesis, and mutating this residue results in diminished activity

in a mini-genome system, suggesting that P may regulate viral

RNA synthesis through its interaction with M2-1 [12]. However,

the role of this phosphorylation site has not been examined in the

context of virus infection. The P protein of HPIV3 is phosphor-

ylated by protein kinase C isoform f (PKC-f) [13] and the serine

residue at position 333 is the likely target site [14]. However, the

role of phosphorylation at Ser 333 in the virus life cycle has not

been reported. Thus, to the best of our knowledge, regulation of

PLoS Pathogens | www.plospathogens.org 1 July 2009 | Volume 5 | Issue 7 | e1000525

paramyxovirus viral gene expression by phosphorylation state of P

has never been directly demonstrated in virus-infected cells even

though it is thought that the phosphorylation of the P protein is

critical for its function in viral gene expression.

PIV5, formerly known as simian virus 5 (SV5) [15], is a

prototypical paramyxovirus of the Rubulavirus genus of the family

Paramyxoviridae [1]. The PIV5 genome encodes seven genes, from

which eight viral proteins are made [1]. The nucleocapsid protein

(NP), phosphoprotein (P) and large RNA polymerase (L) protein

are essential for viral RNA synthesis (mRNA transcription and

genome RNA replication). The V protein plays important roles in

viral pathogenesis. The V/P gene of PIV5 is transcribed into both

the V mRNA and the P mRNA through a process of pseudo-

templated addition of nucleotides, in which the V mRNA is made

by faithful transcription the V/P gene and the P mRNA is made

by co-transcriptional insertion of two non-templated G residues at

a specific site. As a result, the V/P gene is transcribed into two

mRNAs and translated into two proteins, which share identical N-

termini (164 amino acid residues) but different C-termini.

Previously, a strain of PIV5, called canine parainfluenza virus

(CPI+), that causes a neurological disorder in canines was isolated

[16]. During the course of studying the CPI+ virus, a derivative of

CPI+, termed CPI2, was isolated from a dog that was

experimentally infected with the CPI+ virus [17]. Several

differences were found among the genomes of CPI+, CPI2 and

the commonly used lab strain, W3A, including in the shared

region of the V/P gene [15,18,19]. The CPI2 virus has eight

amino acid residues in the V/P gene that are different from the

W3A strain (referred to as wild type in this work) and six of them

are in the shared region of the V and the P proteins (the N-

terminus 164 amino acid residues). The CPI+ virus has five amino

acid residues different from wild type PIV5 in the V/P gene and

three of them are in the shared region of the V and P proteins

(Fig. 1A). A recombinant PIV5 (W3A strain) containing the CPI2

mutations in the shared V and P proteins (rPIV5-CPI2) was

shown to have elevated viral gene expression compared to the wild

type W3A strain and to induce expression of host anti-viral

response genes, such as interferon (IFN)-b and interleukin (IL)-6,

and apoptosis in infected cells [20]. Our recently work indicates

that a single amino acid residue of P, the serine at position 157

(S157), is responsible for the increased viral gene expression. S157

is phosphorylated and its phosphorylation correlates with

decreased viral gene expression as well as the lack of induction

of cytokine expression in infected cells [21], suggesting that

phosphorylation at S157 plays a role in regulating viral gene

expression. However, mechanism of this regulation is not clear.

Intriguingly, S157 residue of P is within a consensus binding

motif for Polo-like kinase 1 (PLK1), a serine/theronine kinase that

plays a critical role in regulating the cell cycle [22]. PLK1 contains

two domains, a polo-box domain (PBD) and a kinase activity

domain. It is known that the PBD domain of PLK1 binds to its

target through a SpS(orT)P motif (SSP motif), where the second

amino acid residue, S or T, is phosphorylated for optimal binding

[23]. The binding of PLK1 to a target protein with an SSP motif

enables PLK1 to phosphorylate either the target itself or a protein

associated with the target [24]. Here, we have investigated the role

of PLK1 in the increased viral gene expression of rPIV5-CPI2.

Results

PLK1 interacts with P in both infected and transfectedcells via the SSP motif

Previously, it has been reported that rPIV5-CPI+, a recombinant

virus containing three amino acid changes (V32I, T33I, and S157F)

compared to wild type has elevated viral gene expression levels and

that S157F is responsible for the phenotype [21]. To ascertain the

mechanism of the increased viral gene expression, the amino acid

sequence of P was searched for known motifs. S157 is in a PLK1

binding motif [23,25], suggesting that PLK1 may play a role in

regulating viral gene expression. To investigate a possible role of

PLK1 in viral gene expression, we examined the interaction

between PLK1 and P. HeLa cells were mock-infected or infected

with PIV5 or rPIV5-CPI+. The cell lysates were immunoprecip-

itated with anti-P antibody and then immunoblotted with anti-P or

anti-PLK1 antibody. As shown in Fig. 1B, PIV5 P protein

interacted with endogenous PLK1, while P protein of rPIV5-

CPI+ (Pcpi+) did not. Since P interacts with viral proteins NP and L,

we wanted to determine whether PLK1 interacts directly with P.

Therefore, we examined the interaction between PLK1 and P in

transfected cells. As shown in Fig. 1C, PLK1 was co-immunopre-

cipitated with P, but not with Pcpi+ by antibody against P; the

reciprocal immunoprecipitation with anti-Flag antibody confirmed

the interaction of PLK1 with P and not Pcpi+, indicating that PLK1

interacts with P, likely through the SSP motif centered at the residue

S157. Furthermore, we carried out the same experiments using

immunoprecipitation followed by immunoblotting and obtained the

same results (supplemental Fig. S1).

PLK1 down-regulates PIV5, but not rPIV5-CPi+ geneexpression

To investigate the role of PLK1 in PIV5 gene expression, we

tested the effect of a PLK1 inhibitor (BI 2536) on reporter gene

expression using a recombinant PIV5 expressing Renilla luciferase,

(rPIV5-RL). As shown in Fig. 2A, BI 2536 at 0.05 and 1 mM

increased Renilla luciferase activity in HeLa cells. In BSR T7 cells,

BI2536 was also effective in increasing the luciferase activity, albeit

at higher concentrations. Because of concerns over potential off-

target effects of this molecule, we also tested the effect of a

structurally different PLK1 inhibitor, GW843682 (Sigma), in

HeLa cells and found the same effect on reporter gene expression

as observed with BI 2536 (supplemental Fig. S2). Furthermore, we

examined the effect of PLK1 inhibition on viral protein

expression. HeLa cells were infected and then metabolically

labeled with 35S-Cys/Met. The infected cells were immunopre-

cipitated with antibody (Pk) that recognizes both the P and V

protein. Compared with control, BI 2536 treatment increased

PIV5 viral protein expression (Fig. 2B) (NP and L bind to P and

Author Summary

The paramyxoviruses are a family of negative-sense RNAviruses that includes many important human and animalpathogens. Paramyxovirus RNA synthesis requires the viralphosphoprotein (P) and the large (L) protein. Phosphory-lation of P is thought to regulate viral gene expression,though direct proof remains elusive. Here, we show thatPolo-like kinase 1 (PLK1) interacts with the P protein ofparainfluenza virus 5 (PIV5), a prototypical paramyxovirus,through the serine residue at position 157 and phosphor-ylates P at serine residue 308. Inhibition of PLK1 increasesviral gene expression. Increased viral gene expression incells infected by PIV5 with mutations in the PLK1 binding/phosphorylation sites correlates with increased inductionof cell death and cytokine expression, suggesting that PIV5limits its gene expression to avoid induction of innateimmune responses. It is possible that targeting PLK1 willenhance host innate immune responses, leading to a novelstrategy of anti-paramyxovirus therapy.

PLK1 Down-Regulates Virus Gene Expression


thus were co-immunoprecipitated with this antibody). In contrast,

BI 2536 treatment did not greatly affect cellular protein

expression, as shown in the right panel in Fig. 2B, indicating that

PLK1 regulation of gene expression in infected cells is specific to

PIV5. Interestingly, BI 2536 did not increase viral protein

expression in rPIV5-CPI+-infected cells (Fig. 2C), suggesting that

PLK1 does not play a role in regulating viral gene expression of

rPIV5-CPI+. These results suggest that the SSP motif plays a

critical role in the inhibition of viral gene expression by PLK1. To

further examine the role of PLK1 in viral gene expression, we used

a mini-genome system, in which viral gene expression can be

examined free of viral infection [21]. As shown in Fig. 2D, PLK1

inhibitor BI 2536 increased reporter gene activity from the PIV5

mini-genome using wt P, but had little effect on luciferase

expression driven by Pcpi+. This result is consistent with the

observation in infected cells. We also attempted to utilize siRNA

technology to knock down expression of endogenous PLK1 in

HeLa cells (data not shown) to examine the effect PLK1 depletion

on viral gene expression. Unfortunately, the experiment was

inconclusive due to the toxicity of siRNA against PLK1. Optimal

knock down of PLK1 expression occurs between 48 to 96 hours

after transfection of siRNA, after which we infected the siRNA-

treated cells. Since PLK1 plays a critical role in cell cycle

progression, this long period of PLK1 depletion, coupled with viral

infection, is overwhelmingly cytotoxic.

Since phosphorylation of S157 reduces activity of the P protein,

we reasoned that increased phosphorylation of the residue by

PLK1 overexpression would further reduce the activity of P. When

PLK1 was ectopically expressed in the mini-genome system, PLK1

transfection reduced the mini-genome activity in a dose dependent

manner (Fig. 2E): there was about 50% reduction of luciferase

activity using 2 ng, and 95% reduction with 16 ng, of PLK1

plasmid in the transfection. PLK1 overexpression had a much

weaker effect on mini-genome transcription when Pcpi+ was used

in place of wild type P; PLK1 plasmid at 4 ng, 8 ng and 16 ng

moderately inhibited the Pcpi+ mini-genome activity, while there

was no significant effect at 2 ng though expression levels of PLK1

correlated with amount of plasmids transfected. To confirm that

PLK1 binds the SSP motif, we performed similar experiments

using P containing a single point mutation S157A instead of Pcpi+,

which has three point mutations (V32I, T33I, and S157F). As

expected, P-S157A failed to bind PLK1 (supplemental Fig. S3A)

and the mini-genome activity using P-S157A was weakly affected

by PLK1 over-expression (supplemental Fig. S3B). Those data

Figure 1. Interaction between P and PLK1. (A). Amino acid residue differences among N-termini of wild type P, Pcpi2 and Pcpi+. (B). Interactionbetween P and endogenous PLK1 in infected cells. HeLa cells were infected with mock, PIV5 or rPIV5-CPI+ and immunoprecipitated (IP) with anti-Pantibody. The immunoprecipitated proteins were resolved in SDS-PAGE and subjected to immunoblotting (IB) with antibody against PLK1. (C).Interaction between P and PLK1 in transfected cells. BSR T7 cells were transfected with plasmids encoding P or Pcpi+ along with Flag-PLK1. Vector(pCAGGS) was used to keep the total amount of transfected plasmids constant. At 18–20 hours post transfection, the transfected cells weremetabolically labeled with 35S-Cys/Met for 3 h at 37uC. The cells were lysed by RIPA buffer and immunoprecipitated with Pk antibody or Flagantibody. The precipitated proteins were resolved by SDS-PAGE gel and visualized using a PhosphorImager.doi:10.1371/journal.ppat.1000525.g001



Figure 2. Effects of PLK1 inhibitor and over-expression of PLK1 on viral gene expression. (A). Effect of PLK1 inhibitor BI 2536 onrecombinant PIV5 expressing Renilla luciferase (rPIV5-RL). HeLa or BSR T7 cells in 24-well plates were infected with rPIV5-RL at MOI of 1 and incubatedwith BI 2536 at concentration of 0.01, 0.05, 0.1, 0.2, 0.5 and 1 mM. Renilla luciferase activity was measured at 18–20 hours post infection and averageluciferase activity +/2 standard derivation of mean (SEM) is shown. (B). Effect of PLK1 inhibitor BI 2536 on viral gene expression. HeLa cells weremock-infected or infected with PIV5 or rPIV5-CPI+ and incubated in 1 mM BI 2536. The cells were metabolically labeled with 35S-Met/Cys at 18–20 hours after infection and immunoprecipitated with Pk antibody, which recognizes V and P and co-precipitates NP and L. The right panel is labeledcell lysates without immunoprecipitation. (C). Quantification of Figure 2B. Six independent experiments were performed and the relative amount ofthe P protein was measured. The level of PIV5 P in infected cells with DMSO treatment was set at 100 and the rest was normalized to the level of PIV5P. The average of relative expression level of P +/2 SEM is shown. (D). Effects of PLK1 inhibitor on the mini-genome system. A mini-genome plasmid(pSMG-RL) that contains a Renilla luciferase (RL) reporter gene and from which a negative-sense mini-genome is generated from T7 RNA polymerasetranscription in BSR T7 cells was described previously [21]. In the presence of NP and L and P or P from rPIV5-CPI+ (Pcpi+), this negative-sense RNAtemplate is replicated and transcribed to give rise to the RL mRNA, resulting in luciferase activity. The pT7-FF-Luc plasmid, which contains an FF-Lucreporter gene as a transfection efficiency control, was transfected along with the plasmids. Firefly and Renilla luciferase activities were detected in celllysates at 18 to 20 h post-transfection, as described in Experimental Procedures. BSR T7 cells were transfected with necessary plasmids and incubatedwith 0.25 mM BI 2536. At 18–20 hours post transfection, the cells were lysed and dual luciferase assay was performed. The relative luciferase activitiesare calculated at ratios of Renilla luciferase activity (indicative of mini-genome replication) versus FF-Luc activity (indicative of transfection efficiency).The average of relative luciferase activity +/2 SEM is shown. (E). Effects of PLK1 overexpression on the mini-genome system. Plasmids encodingpCAGGS-Flag-PLK1 at various concentrations were transfected along with the mini-genome system as above. Dual luciferase assay was carried asdescribed above. An aliquot of cell lysate was used for immunoblotting to detect expression levels of Flag-PLK1 (the bottom panel).doi:10.1371/journal.ppat.1000525.g002



indicate that PLK1 binds to PIV5 P protein through the SSP motif

centered at the residue S157 and reduces PIV5 gene expression.

The moderate inhibitory effect of over-expressed PLK1 on Pcpi+or P-S157A mini-genome system is likely due to an interaction

between PLK1 and an additional SS(T)P sequence centered at

T108 of P. This STP sequence is not known to be phosphorylated

and mutating this T to A had no effect on PLK1 binding or viral

gene expression (data not shown). However, since even unpho-

sphorylated STP can weakly interact with PLK1, overexpression

of PLK1 may result in interaction of PLK1 (a low level of Pcpi+and PLK1 was indeed detected when PLK1 was overexpressed in

supplemental Fig. S1), resulting in the moderate inhibitory effect.

Kinase activity of PLK1 is required for its inhibitory effecton PIV5 gene expression

To examine whether the binding of PLK1 is sufficient for its

inhibitory effect on PIV5 gene expression, or if its kinase activity is

required, we tested a the effect of overexpression of a kinase-

deficient PLK1 (PLK1-K82M), which has a lethal point mutation

at the ATP binding site (K82M), on PIV5 minigenome activity

[26]. As shown in Fig. 3A, PLK1-K82M, at amounts from 2 to

16 ng, failed to inhibit PIV5 mini-genome activity, in contrast to

wt PLK1, even though PLK1 and PLK1-K82M were expressed at

similar levels, indicating that kinase activity is required for PLK1’s

inhibitory effect on PIV5 gene expression. Since the binding

between PLK1 and target protein is through the PBD domain at

the C-terminus of PLK1, we expected that PLK1-K82M would

still bind the P protein and this was confirmed by co-

immunoprecipitation (Fig. 3B). Thus, the kinase activity of

PLK1 is required for PLK1’s effect on PIV5 gene expression,

while the binding itself is not sufficient.

PLK1 phosphorylates the P protein in infected cellsTo investigate the mechanism by which PLK1 regulates PIV5

gene expression by PLK1, it was important to determine the

phosphorylation target of PLK1. PLK1 can phosphorylate SSP

motif-containing proteins themselves or those proteins associated

with the SSP-containing protein. We hypothesized that after

binding to the P protein, PLK1 phosphorylates P itself in infected

cells. To test this, we compared the level of phosphorylated P

protein with or without PLK1 inhibitor BI 2536 treatment in PIV5

or rPIV5-CPI+ infected cells. Consistent with previous studies

[21], rPIV5-CPI+ infected cells showed higher levels of viral

protein expression than that PIV5 infected cells and Pcpi+ had

lower level of phosphorylation than that of PIV5 (Fig. 4A). Because

we only treated the infected cells for a short period of time during

labeling (4 hours), no significant difference of PIV5 viral protein

levels between DMSO and BI 2536 treated cells was detected by35S-Cys/Met labeling. However, there was a significant reduction

of phosphorylation of P due to BI 2536 treatment in PIV5 infected

cells (Fig. 4B), but not in rPIV5-CPI+ infected cells. Interestingly,

while NP was also phosphorylated in infected cells, its phosphor-

ylation was not affected by PLK1 inhibitor treatment, indicating

that NP phosphorylation is PLK1-independent even though it

interacts with P.

PLK1 phosphorylates the P protein at S308Because PLK1 inhibition reduced P phosphorylation, we

wanted to determine the PLK1 phosphorylation site within P.

Since a consensus PLK1 phosphorylation site has been reported

[27], we searched the sequence of P and found that the serine

residue at position 308 (S308) within P protein resembles a PLK1

phosphorylation site (Fig. 5A). Mutation of S308 should thus result

in a P with higher activity in the mini-genome system that is

insensitive to PLK1. As shown in Fig. 5B, mutating S to A at 308

(P-S308A) increased the mini-genome activity when compared

with wild type P. Mutating S304 and S313, two other S residues

close to S308, had no effect on the mini-genome activity

(supplemental Fig. S4A and S4B). In addition, PLK1 inhibitor

BI 2536 treatment did not affect P-S308A mini-genome activity

(Fig. 5C). Furthermore, overexpression of PLK1 had no effect on

the activity of the mini-genome system using P-S308A (Fig. 5D),

Figure 3. Effect of kinase-deficient PLK1 on PIV5 protein expression. (A). Effects of overexpression of PLK1 kinase-dead mutant K82M on themini-genome system. BSR T7 cells in 24-well plates were transfected with various concentrations of pCAGGS-Flag-PLK1 or pCAGGS-Flag-PLK1-K82Mtogether with the plasmids necessary for the PIV5 mini-genome system. Dual luciferase assay was carried out after 18–20 h post transfection asbefore. NC: negative control, no plasmid encoding P; PC: positive control, no plasmid encoding PLK1 or PLK1 K82M. The relative activity unit ofpositive control was set as 100. The average of relative luciferase activity +/2 SEM is shown. An aliquot of cell lysates was used for immunoblotting todetect the expression level of Flag-PLK1 or Flag-PLK1-K82M. (B). Interaction between P and PLK1-K82M. A plasmid encoding P was transfected intocells with a plasmid encoding Flag-PLK1 or Flag-PLK1-K82M. The cells were metabolically labeled and immunoprecipitated with anti-P or anti-Flag.doi:10.1371/journal.ppat.1000525.g003



indicating that S308 is a PLK1 phosphorylation target. To exclude

the possibility that the mutation at S308 caused a conformational

change in P such that it can no long be bound by PLK1, we

examined the interaction between PLK1 and P-S308A. As shown

in Fig. 5E, P-S308A co-precipitated with PLK1 and vice versa.

To confirm that PLK1 phosphorylates P protein at S308, we

carried out an in vitro kinase assay. P proteins were purified from

HeLa cells transfected with plasmids encoding either P or P-

S308A by affinity chromatography using anti-P antibody conju-

gated agarose gel. Purified P was then incubated with PLK1

purchased from a commercial vendor. As shown in Fig. 5F, wild

type P was phosphorylated by PLK1, while P-S308A was not,

indicating that S308 within P protein is the phosphorylation site of

PLK1.

PLK1 targets S308 of P in infected cellsPreviously, mutations affecting P phosphorylation were identi-

fied and tested using in vitro or mini-genome assays. However, the

results were not substantiated when the sites are incorporated into

recombinant viruses. Thus, to determine whether phosphorylation

of P affects viral gene expression, it is essential to examine the

effects of putative mutations in the context of virus infection. Using

a PIV5 reverse genetics system, we generated two recombinant

viruses encoding mutant P proteins: rPIV5-V/P-S157A, which has

a single amino acid change in the PLK1 binding site; and rPIV5-

P-S308A, which has a single amino acid residue change at 308 at

the PLK1 phosphorylation site (Fig. 6A). The mutant viruses grew

to normal titers and formed plaques similarly to wt PIV5 (data not

shown). Viral gene expression levels from cells infected with the

viruses were examined by flow cytometry using anti-HN antibody.

As shown in Fig. 6B, cells infected by rPIV5-V/P-S157A or

rPIV5-P-S308A expressed higher levels of HN than PIV5-infected

cells, consistent with previous observations.

To investigate whether PLK1 phosphorylates the P protein at

S308 in infected cells, we performed 33P-orthophosphate labeling

of infected cells with or without BI 2536 treatment similar to the

experiment described in Fig. 4A. As shown in Fig. 6C and 6D, the

P protein in cells infected by rPIV5-P-S308A had lower levels of

phosphorylation than that in PIV5-infected cells. Phosphorylation

of P-S308A in infected cells was not sensitive to BI 2536 treatment

while phosphorylation level of wild type P was reduced by PLK1

inhibitor BI 2536, indicating that S308 is a phosphorylation target

for PLK1 in infected cells. These results mirror those presented in

Fig. 4 using rPIV5-Cpi+, whose P protein encodes Ser at position

308 but no longer binds PLK1 due to the S157F mutation,

underscoring the importance of both binding and phosphorylation

of P by PLK1.

Induction of apoptosis and cytokine expression bymutant viruses

It has been reported that rPIV5-CPI2 virus, which contains a

mutation at S157, causes increased cell death and cytokine

expression in addition to the increased levels of viral gene

expression. To investigate whether recombinant PIV5 containing

these mutations described above cause increased cell death, MDBK

cells were infected and photographed. As shown in Fig. 7A, the

viruses caused increased cell death. To further investigate the nature

of the cell death, apoptosis assays were performed on the cells

infected with mock, wild type, rPIV5-P-S308A or rPIV5-V/P-

S157A. Both mutant viruses induced increased apoptosis in the

infected cells using propidium iodine staining, which detects the sub-

G0-G1 population, annexin V staining, which detects phosphati-

dylserine (PS) on the surface, and TUNEL assay, which detects

nicked DNA (Fig. 7B, C and D). To examine growth of

recombinant viruses in cultured cells, HeLa cells and Vero cells

were infected with recombinant viruses at low MOI and virus titers

in media of infected cells were determined at various times post-

infection by plaque assay. Interestingly, while rPIV5-V/P-S157A

and rPIV5-P-S308A grew fast than wild type initially as expected

since the mutant viruses had higher levels of viral gene expression,

their growth was reduced at later time points in HeLa cells, which

are capable of producing and responding to interferon. This drop of

growth was not observed in Vero cells, which do not produce

interferon (Fig. 7E), indicating that interferon may play a role in

growth of the viruses. To investigate induction of cytokines by the

mutant viruses, HeLa cells were infected and media from infected

cells were collected. Amounts of IFN-b, an anti-viral cytokine, and

IL-6, a proinflammatory cytokine, were measured using ELISA.

Expression levels of both cytokines were elevated in rPIV5-P-S308A

and rPIV5-V/P-S157A-infected cells compared to mock or wild

type virus-infected cells (Fig. 7F and G).

Discussion

Our studies present the first evidence that phosphorylation of a

paramyxovirus P protein directly regulates viral gene expression.

Mutating phosphorylation sites within the P protein of SeV such

that P phosphorylation is reduced by more than 90% in

Figure 4. Effects of PLK1 inhibitor on phosphorylation of P. (A). Effects of BI 2536 on phosphorylation of P. Cells were mock-infected orinfected with PIV5 or rPIV5-CPI+. At 18–20 hours post infection, the cells were metabolically labeled with 35S-Cys/Met or 33P-orthophosphate in thepresence of 1 mM PLK1 inhibitor BI 2536 or DMSO as described in Experimental Procedures. The cells were lysed and immunoprecipitated with anti-PPk antibody. (B). Quantification of effects of BI 2536 on phosphorylation of P. Three individual experiments as described in (A) were quantified. Theaverage ratio of phosphorylated P from the 33P-labeling experiment to total amount of P as 35S-labeled P in PIV5 infected cells treated with DMSOwas set to 100 and the others are normalized to the ratio. The average of relative level of phosphorylation of P +/2 SEM is shown.doi:10.1371/journal.ppat.1000525.g004



transfected cells does not affect its activity in transcription, either

in a mini-genome system or in recombinant virus [5]. Similarly,

mutation of the major phosphorylation sites within the P protein of

RSV results in a mutant P whose level of phosphorylation is

reduced by more than 90%. Yet, when these mutations are

introduced into the RSV genome by reverse genetics, expression

levels of the viral genes are not adversely affected. These results

seemingly suggest that phosphorylation of the P proteins of

paramyxoviruses does not have a role in viral gene expression.

However, it is possible that the critical phosphorylation sites within

the P proteins have not been identified in these earlier studies.

While it is true that some of the phosphorylation sites may be

superfluous and may not have a role in viral gene expression, it is

difficult to imagine that none of the phosphorylation sites within a

heavily phosphorylated P protein have any role in its function.

Even though mutation of known phosphorylation sites reduces

phosphorylation of SeV and RSV P by 90%, it is possible that the

remaining residues that are phosphorylated (even though they only

count for about 10% of total phosphorylation) are critical for viral

gene expression. In this work, we have identified a host kinase,

PLK1, which phosphorylates the P protein and mapped both its

binding (SSP motif at S157) and phosphorylation (S308) sites

within P. Further, we demonstrated that this site (S308) is

important in regulating viral gene expression using a recombinant

virus. To the best of our knowledge, this is the first report of a host

kinase that directly phosphorylates the P protein and regulates

viral gene expression of a paramyxovirus. Interestingly, PLK1

phosphorylation down-regulates viral gene expression, contrary to

our expectation that phosphorylation may be essential for viral

gene expression. It is possible that additional phosphorylation sites

within the P protein may play an essential role in the regulation of

viral gene expression. Previously, we have found that AKT1, a

serine/threonine kinase, plays a critical role in viral gene

expression [28]. However, direct interaction between P and

AKT1 as well as phosphorylation site of AKT within the P protein

has not been reported. Further studies of phosphorylation of the P

protein will be needed to identify possible phosphorylation sites

within P that are critical for viral gene expression.

Previously, we showed that S157 was phosphorylated and

contributes to the phenotypes of rPIV5-CPI2. However, it was

Figure 5. PLK1 phosphorylates the P protein at S308. (A). Prediction of PLK1 phosphorylation site within P protein. The center bold S/T residueis the PLK1 phosphorylation site. X, any amino acid residue; Y, amino acid residue with a hydrophobic side chain. (B). Effect of mutating S308 into Aon the mini-genome activity. A mini-genome system as described before was used. To ensure a validity comparison of P vs. P-S308A, a range ofconcentrations of the P proteins in the experiments were used. Cell lysate aliquots of the transfected cells were subjected to immunoblotting usinganti-NP or anti-P antibody. The average of relative luciferase activity +/2 SEM is shown. (C). Effects of PLK1 inhibitor on the mini-genome systemusing P-S308A. To study the effect of PLK1 inhibitor on mini-genome activity with mutant P-S308A, BI 2536 was added to the mini-genome systemusing P-S308A in a similar fashion as in Fig. 2D. (D). Effects of PLK1 overexpression on the mini-genome system using P-S308A. Plasmids encodingpCAGGS-Flag-PLK1 at various concentrations were transfected along with the mini-genome system as above. Dual luciferase assay was carried asdescribed above. An aliquot of cell lysate was used for immunoblotting to detect expression levels of Flag-PLK1. (E). Interaction between P-S308A andPLK1. A plasmid encoding Flag-PLK1 was transfected into cells with a plasmid encoding P or P-S308A. The cells were metabolically labeled andimmunoprecipitated with anti-P or anti-Flag. (F). Phosphorylation of P by PLK1 in vitro. HeLa cells were transfected with plasmids pCAGGS-P orpCAGGS-P-S308A. P and P-S308A were purified from transfected cells using anti-P as described in Experimental Procedures. Kinase assays werecarried out in a total volume of 30 ml containing 100 ng PLK1 for 60 minutes at room temperature. The bottom panel is the input of P and P-S308Awith PLK1 on a SDS-PAGE with Coomassie blue staining.doi:10.1371/journal.ppat.1000525.g005



formally possible that the V protein, which also contains the S157

mutation, contributed to the CPI2 phenotypes since V and P are

identical in the first 164 amino acid residues. In this work, we have

identified a single amino acid residue (S308) within the P protein

as a PLK1 phosphorylation site. Mutating this phosphorylation

site has the same effect on viral gene expression as the CPI2 P

mutations, further confirming our previous report that the

increased gene expression from rPIV5-CPI2 is due to a P protein

that is more efficient in facilitating viral RNA synthesis since the

S308A mutation is within the unique P sequences while the V

remains intact. It may seem counterintuitive that a virus would

down-regulate its own gene expression by maintaining binding

and phosphorylation sites for PLK1. However, higher PIV5 gene

expression is associated with the induction of cytokine expression

and cell death, which in turn limits virus replication and spread.

Therefore, viruses that down-regulate their gene expression via

PLK1 phosphorylation of P would have advantages in viral

transmission over viruses with defective PLK1 binding or

phosphorylation sites (Fig. 8). Since increased cell death and

cytokine expression induced by virus infection limit virus

replication, it is possible that targeting PLK1 can enhance host

innate immune responses, leading to a novel strategy to control

virus infection. The lower viral gene expression after PLK1

phosphorylation may also favor persistent viral infection. Interest-

ingly, about half the strains of PIV5 have an S residue at position

157 [15].

PLK1 can phosphorylate proteins with a PLK1 binding domain

as well as a proteins associated with PLK1-binding proteins. Thus,

it is possible that PLK1 down-regulates viral gene expression

through phosphorylation of a P-binding protein. However, the

S308A mutant, which still binds PLK1, has the same viral gene

expression phenotype as PLK1 binding site (S157) mutants,

indicating that P is not merely a scaffold for PLK1 to down-

regulate viral gene expression, but is the target of PLK1 itself. The

exact mechanism of how the phosphorylation by PLK1 at residue

S308 affects the function of P remains to be determined. The P

protein does not have a catalytic domain but it plays an essential

regulatory role in viral gene synthesis. P is known to interact with

viral proteins NP and L. Binding to and phosphorylation by PLK1

do not appear to have an impact on the interactions between P

and NP or L. Previously, it has been reported that rPIV5-CPI2

has an increased viral RNA genome synthesis while viral mRNA

transcription is not affected [21]. It is possible that phosphoryla-

tion at S308 changes the conformation of P to make it less efficient

in facilitating viral RNA genome synthesis. It is possible that

phosphorylated P (at 308) binds to RNA better and increases the

rate of RNA synthesis. However, we cannot rule out the possibility

that phosphorylation of P at S308 may affect its ability to interact

with a yet-to-be identified host protein, which in turn regulates

viral gene expression.

A role for PLK1 in infection by other viruses has been reported

recently. Expression of human T-lymphotropic virus type-I

protein p30 reduces expression as well as phosphorylation of

PLK1 [29] while PLK1 is up-regulated in CD4+ T cells from

rhesus macaques infected with simian immunodeficiency virus

[30] as well as in cells expressing E6/E7 of human papillomavirus

Figure 6. Effect of mutating S to A at position 308 in recombinant virus. (A). Sequences of PIV5, rPIV5-V/P-S157A and rPIV5-P-S308A atregions around residue 157 and 308 of the P protein. The entire genome of the mutant viruses was sequenced and no mutation was observed exceptin the designated site (S157A or S308A). (B). Viral gene expression from PIV5, rPIV5-V/P-S157A and rPIV5-P-S308A-infected cells. HeLa cells wereinfected with PIV5, rPIV5-V/P-S157A, or rPIV5-P-S308A. At 16 hours post infection, the infected cells were processed for flow cytometry using anti-HNantibody as described in Experimental Procedures. Average mean fluorescence intensity of infected cells +/2 SEM is graphed. (C). Effects of PLK1inhibitor on phosphorylation of P of rPIV5-P-S308A. Cells were mock-infected or infected with PIV5 or rPIV5-P-S308A. At 18–20 hours post infection,the cells were metabolically labeled with 35S-Cys/Met or 33P-orthophosphate in the presence of 1 mM PLK1 inhibitor BI 2536 or DMSO as described inFig. 4A. The cells were lysed and immunoprecipitated with ant-P Pk antibody. (D). Quantification of effects of BI 2536 on phosphorylation of P ofrPIV5-P-S308A. Three individual experiments as described in (C) were quantified. The average ratio of phosphorylated P from the 33P-labelingexperiment to total amount of P as 35S-labeled P in PIV5 infected cells treated with DMSO was set to 100 and the others are normalized to the ratio.The average of relative phosphorylation level of P +/2 SEM is shown.doi:10.1371/journal.ppat.1000525.g006



Figure 7. Induction of apoptosis and cytokine expression by mutant viruses. (A) Induction of cytopathic effect by virus infection. MDBKcells were infected with PIV5, rPIV5-V/P-S157A, or rPIV5-P-S308A. At 48 hours post infection, the infected cells were photographed using microscope(Nikon ECLIPSE TE300, Japan). (B) Propidium iodine (PI) staining. MDBK cells were infected and collected 2 days post infection as above. The cells werestained with PI and cellular DNA profiles were examined using flow cytometry. Percentages of sub G0-G1 population, which is considered apoptotic,are graphed. S157A, rPIV5-V/P-S157A; S308A, rPIV5-P-S308A. (C). Annexin V staining. The infected cells were stained with FITC-labeled annexin V andmeasured using a flow cytometer. Annexin V, which binds to phosphatidylserine (PS), is an indication of cells undergoing apoptosis when it is presenton cell surface. Percentages of annexin V positive cells are graphed. (D) TUNEL assay. The infected cells were subjected to a TUNEL assay as describedin Experimental Procedures. TUNEL positive cells were measured using a flow cytometer and percentages of TUNEL positive cells are graphed. (E).Growth of recombinant viruses. HeLa cells or Vero cells were infected with viruses at 0.01 MOI and media of infected cells were collected at indicatedtime points. Titers of viruses were determined using plaque assay. (F). Induction of IFN-b. HeLa cells were infected and levels of IFN-b were measuredusing ELISA at 2 days post infection. (G). Induction of IL-6. HeLa cells were infected and levels of IL-6 were measured using ELISA at 2 days postinfection. Error bars are standard deviation of mean.doi:10.1371/journal.ppat.1000525.g007



[31]. PLK1 can directly phosphorylate pp65, a dispensable gene of

human cytomegalovirus [32]. However, none of these studies has

demonstrated a direct role of PLK1 in viral gene expression and

the biological consequence of these studies are not clear. Recently,

Ludlow et al reported that the V protein of NiV contains a PLK1

binding site within the shared region of the NiV P protein and that

the PLK1 binding site is also involved in interacting with Stat1

protein, a key protein in IFN signaling [33]. However, the

biological consequence of this PLK1 interaction is not clear; since

Nipah virus is a biosafety level 4 pathogen, examining the role of

PLK1 in the context of Nipah virus infection is difficult. Ours is

the first report of PLK1 playing a direct role in virus replication.

The PLK1 binding site within the P protein of PIV5 is within

the shared region of the V and P protein; thus, the PIV5 V should

bind PLK1 as well. It is well known that the V protein blocks

interferon signaling by causing degradation of Stat1. However, it is

unlikely this PLK1 binding site is involved in interferon signaling

since the mutations at the S157 residue have no effect on IFN

signaling [19]. It is possible that PLK1 plays additional roles in

viral life cycle. For instance, PIV5 V protein is known to slow

down the cell cycle [34], so it is tempting to hypothesize that PLK1

is involved in the regulation of cell cycle by the V protein.

A cursory examination of the P proteins of paramyxoviruses

indicates that most contain SSP (or STP) sequence motifs; some of the

P proteins have more than one SSP sequence motif. For instance,

NiV P has two SSP sequences. However, it is not known whether

those S/T residues are phosphorylated. Since efficient PLK1 binding

requires the middle S/T residue to be phosphorylated [25], it is not

clear whether these SSP sequences are truly SSP motifs and bind

PLK1. It is possible that the role of PLK1 in viral gene expression is

universal for all paramyxoviruses. On the other hand, it is equally

possible that PLK1 plays different roles for different paramyxoviruses.

These questions warrant further investigation.

Materials and Methods

Plasmids, viruses, and cellsPlasmids expressing P mutants, Pcpi+, P-S157A and P-S308A

were made from a copy of P of W3A strain in the pCAGGS vector

[35]. The human PLK1 gene was obtained from Open Biosystems

(AL, USA). A Flag tag was added to the N-terminus of PLK1 and

cloned into pCAGGS vector. Kinase-deficient PLK1 (Flag-PLK1-

K82M) was constructed by PCR using Flag-PLK1 as the template

and pfx50TM DNA polymerase (Invitrogen, 12355-012), and

confirmed by sequencing. Plasmids containing full-length genome

for rPIV5-V/P-S157A and rPIV5-P-S308A viruses were made

similarly to that of rPIV5-CPI+ as described before [21] and

confirmed by sequencing. Plasmids used in PIV5 mini-genome

system have been described before [21,36]. The details of plasmid

construction and sequences of the plasmids are available on

request. rPIV5-V/P-S157A and rPIV5-P-S308A were rescued

from the plasmids containing their respectively full-length genome

as described before [21]. PIV5, rPIV5-CPI+ and rPIV5-RL

viruses used in this study were described as before [21]. HeLa and

MDBK cells were grown in Dulbecco modified Eagle medium

(DMEM) (Invitrogen, USA) containing 10% fetal bovine serum

(FBS), 100 IU/ml penicillin, and 100 mg/ml streptomycin. BSR

T7 cells were grown in the same media as HeLa plus 10% tryptose

phosphate broth (TPB) and 400 mg/ml G418. All cell lines were

incubated at 37uC in 5% CO2. The growth medium for infected

cells contains only 2% FBS. Virus titers were determined using

BHK cells as described before [37,38].

Figure 8. A working model for regulation of PIV5 viral gene expression by PLK1. For wild type PIV5, PLK1 binds to SSP motif (156–158)within the P protein, and then phosphorylates the P protein at S308, resulting in inhibition of PIV5 gene expression. The reduced viral geneexpression supports efficient virus replication, but enables virus to avoid the induction of cell death and cytokine expression. For rPIV5-CPI+, SSP ismutated to SFP, preventing the binding of P to PLK1 and, thus, no phosphorylation of S308 by PLK1. As a result, viral gene expression is higher thanthat of PIV5, and cell death as well as cytokine production is induced, resulting in limiting replication and spread of virus. However, when PLK1 is overexpressed, PLK1 can interact with S157F weakly through STP centered at T108 and reduces P-S157F mini-genome expression moderately. For rPIV5-P-S308A, while the SSP motif is intact and P-S308A still binds to PLK1; however, there is no phosphorylation of S308. This mutant virus, similar to rPIV5-CPI+, shows elevated viral gene expression and increased induction of cell death.doi:10.1371/journal.ppat.1000525.g008



Viral genome sequencingMDBK cells were infected with mutant virus stock with MOI of

1. At two days post infection, viral RNA was extracted from the

media by QIAmp viral RNA mini kit (Qiagen, CA) following the

protocol provided by the manufacture. The viral RNA was reverse

transcribed with random primer (Superscript III first-Strand

Synthesis System, Invotrogen). 12 pairs of oligomers were used

for PCR and 30 oligomers were used for sequencing. The terminal

sequences of viral genome were obtained by Rapid Amplification

of cDNA Ends (RACE) as described before [39]. Briefly, to

sequence 39 leader sequence, viral RNA was ligated with an

adaptor JX129 then reverse transcribed by JX130 (complementary

to JX129). The cDNA were amplified by the JX130–JX131 pair

and then sequenced. To sequence 59 trailer sequence, the viral

RNA was firstly reverse transcribed using JX133. The cDNA

products were purified and ligated to JX129. The ligation product

was amplified with JX130–JX132. The PCR products were

purified, sequenced and compared with wide type PIV5 genome

using BLAST. Sequences of all oligomers used for PCR or

sequencing are available on request.

ImmunoprecipitationTo detect interactions between P and endogenous PLK1 in

infected cells, HeLa cells in 10 cm plates were infected with mock,

PIV5 or rPIV5-CPI+ at multiplicity of infection (MOI) of 3 for 12 h

then treated with 100 ng/ml Nocodazole, which is known to increase

expression levels of PLK1 by arresting cells into M phase when PLK1

expression levels are the highest [40–42]. After10–12 h incubation

with Nocodazole, the cells were lysed by Whole Cell Extraction

Buffer (WCEB: 60 mM Tris-HCl pH 6.8, 40% glycerol, 4%SDS,

3% dithiothreitol (DTT), and few grains of bromophenol blue)

containing protease inhibitor cocktail as described before [21,43].

The same amount of the lysates were pre-cleared with 40 ml protein

G sepharose beads for 1 h at 4uC. The cell lysate were then incubated

with P antibody and 30 ml protein G sepharose beads for 2–3 h at

4uC. The precipitated proteins were resolved by 15% sodium dodecyl

sulfate polyacrylamide gel electrophoresis (SDS-PAGE), and trans-

ferred onto PVDF membrane (Milipore, MA). Mouse anti-PLK1

antibody (Santa Cruz Biothechnology) was used to detect PLK1 by

immunoblotting as described before [21,43]. To examine whether

PLK1 interacts with P directly, not through other viral proteins, BSR

T7 cells in 6 cm plates were transfected with pCAGGS vector,

pCAGGS-P, or pCAGGS-Pcpi+ with pCAGGS-Flag-PLK1 using

Plus and Lipofectamine reagents following the protocol from

manufacturer (Invitrogen). The transfected cells were incubated with

DMEM deficient of Cys/Met for 30 minutes and then labeled with

DMEM containing 35S-Cys/Met Promix (Amersham Life Sciences)

(100 mCi/ml) for 3 h. The cells were lysed in RIPA buffer (20 mM

Tris-HCl pH 7.4, 150 mM NaCl, 0.2% Triton-X100, 0.1% SDS,

5 mM Iodoacetanide) containing protease inhibitor cocktail. Pk

antibody or Flag antibody together with Protein G sepharose beads

were used for immunoprecipitation. The precipitated proteins were

washed three times with RIPA buffer containing 0.3 M NaCl and

once with RIPA buffer containing 0.15 M NaCl. After washing, the

precipitated proteins were resolved by 15% SDS-PAGE, and proteins

were detected by autoradiography using Storm Phosphorimager

(Molecular Dynamics Inc., Sunnyvale, CA).

PLK1 inhibitorsBI 2536, a highly selective PLK1 inhibitor was purchased from

Axon Company (The Netherlands) [41]. The compound was

dissolved in dimethyl sulfoxide (DMSO). To study the effect of

PLK1 inhibitor on rPIV5-RL gene expression, HeLa cells or BSR

T7 cells at about 90% confluence were infected with rPIV5-RL at

a MOI of 1 and incubated with the media containing 2% FBS and

PLK1 inhibitor. Renilla luciferase activity was examined by

luminometer using Renilla luciferase assay kit (Promega) at 16 to

20 hours post infection. Similarly, BI 2536 was tested in the mini-

genome system. GW843682 (Sigma), a PLK1 inhibitor, which

inhibits PLK1 and weakly inhibits PLK3 (which has similar

function as PLK1) [44], was also used.

Phosphorylation of PTo examine phosphorylation of P, infected cells in 6 cm plates

were incubated in the medium lacking sodium phosphate for

30 minutes and then labeled with medium containing 33P-

orthophosphate (100 mCi) in the present of DMSO or 1 mM BI

2536 for 4 hour. The cells were then lysed, immunoprecipitated

using Pk antibody, and resolved in 15% SDS-PAGE and

quantified as before.

PIV5 mini-genome system and dual luciferase assayPIV5 mini-genome system used in this study was described

before [21], except that negative controls contained no P plasmid,

instead of no L plasmid. BSR T7 cells in 24-well plates were

transfected with plasmids expressing pSMG-RL (0.2 mg), NP

(0.2 mg), P (0.01–0.16 mg), L (0.3 mg) as well as a firefly luciferase

gene (FF-Luc) under control of a T7 promoter using Plus and

Lipofectamine reagents (Invitrogen). GFP plasmid was used as a

control to keep the total amount of transfected plasmids the same.

For transfection of Flag-PLK1 or Flag-PLK1-K82M together with

the mini-genome system composed of P, P-S157A orP-S308A, P

or P mutant plasmids were used at 0.02 mg. At 18–22 hrs post

transfection, the cells were lysed with 100 ml 16passive lysis buffer

(Promega). 10 ml lysate from each well were used for dual

luciferase assay by dual luciferase assay kit (Promega). Ratios of R-

Luc to FF-Luc activity are normalized as relative activity unit. Six

replicates of each sample were used for statistical analysis. An

aliquot of the cell lysate from mini-genome system was mixed with

equal volume of 26 protein lysis buffer as described before.

Samples were resolved in 10% SDS-PAGE and transferred onto

PVDF membrane. Mouse anti-NP and mouse anti-P (Pk antibody)

were used together for immunoblotting. Anti-Flag was used to

detect the amount of input of Flag-PLK1 or Flag-PLK1-K82M.

In vitro kinase assayFor PLK1 in vitro kinase assay, HeLa cells in 10 cm plates were

transfected with pCAGGS-P-wt or pCAGGS-P-S308A. At

24 hours post transfection, anti-V5 agarose gel, which has Pk

antibody conjugated to the agarose beads, was used to purify P-wt

or P-S308A using immunoprecipitation. Similar amounts of P-wt

and P-S308A were used for PLK1 (Cell Signal, MA) in vitro kinase

assay with kinase buffer (25mMHEPES, 25 mM beta-glycero-

phosphate, 25 mM MgCl2, 2 mM dithiothreitole, and 0.1 mN

NaVO3). Half of the reaction mixture was used for SDS-PAGE

followed by Coomassie blue staining to measure the amount of

input P-wt and P-S308A, and the other half of the mixture was

incubated with 10 mCi c-32P-ATP (Amersham Biosciences) for 1 h

at room temperature in a total volume of 30 ul containing 100 ng

PLK1. Reactions were terminated by addition of the same volume

of 26SDS loading buffer. Proteins were separated by 15% SDS-

PAGE and phosphorylation was detected with a Storm Phosphor-

Imager (Molecular Dynamics).

Flow cytometryTo detect different levels of viral gene expression in Mock,

PIV5, rPIV5-P-S308A, and rPIV5-V/P-S157A infected cells, flow



cytometry was carried out as previously descrbed [21]. Briefly,

HeLa cells at about 90% confluence were mock infected or PIV5,

rPIV5-V/P-S157A, or rPIV5-P-S308A infected at MOI of 3. The

infected cells were collected at 16 h hpi and fixed with 0.5%

formaldyhyde for 2 h at 4uC. After centrifugation and removal of

the supernatant, the cells were re-suspended in 500 ml solution of

FBS-DMEM (50:50), then permeabilized in 70% ethanol

overnight at 4uC. The cells were washed with phosphate buffered

saline without Mg2+ and Ca2+ (PBS-) and incubated with mouse

anti-HN in PBS- with 10% FBS for 30 min at room temperature.

After washing with PBS-, the cells were further stained with FITC-

conjugated goat anti-mouse antibody in the dark and washed with

PBS-. The fluorescence intensity was measured using a flow

cytometer (FC500).

Apoptosis assaysTo detect apoptotic cells, annexin-V staining, propidium iodide

(PI) staining, and terminal deoxynucletidyltransferase-mediated

dUTP-FITC nick end labeling (TUNEL) assay were performed as

previously described [45,46]. Briefly, MDBK cells in 6-well plates

were infected with Mock, PIV5, rPIV50P-S308A or rPIV5-V/P-

S157A with MOI of 5 for 48 h. The floating cells were harvested

together with the trypsinized monolayer cells. For annexin-V

staining, the cells were washed with cold PBS- and incubated with

FITC-labeled annexin-V (Annexin-V-FLIOS, Roche Diagnostics

Corp. Mannheim, Germany) for 15 min at room temperature.

10,000 cells were examined for fluorescence by a flow cytometer

FC500. For PI staining, the collected cells were fixed and

permeabilized as before. The permeabilized cells were washed

and incubated with 100 ml Pk antibody (1:100 dilution in PBS-1%

BSA) for 30 min at room temperature. After washing, the cells

were incubated with 100 ml FITC-labeled anti-mouse antibody

(1:1,000 in PBS-1%BSA). The cells were finally incubated with

500 ml of PI solution (50 mg/ml) (Sigma-Aldrich). The stained cells

were examined with a flow cytometer. For TUNEL assay, the

permeabilized cells were incubated with 25 ml of TUNEL reaction

mixture (Roche Diagnostics Corp. Mannheim, Germany) for 3 h

in the dark at 37uC. After washing, the cells were incubated with

100 ml Pk antibody (1:100 dilution in PBS-1% BSA) then 100 ml

Phycoerythrin-labeled anti-mouse antibody (1:100 dilution in

PBS-1%BSA) and analyzed with FC500.

Enzyme-linked immunosorbent assay (ELISA) for IFN-band IL-6

HeLa cells infected with mock, PIV5, rPIV5-P-S308A, rPIV5-

V/P-S157A were infected with MOI of 5. The media were

collected at 48 h post infection and centrifuged to remove cell

debris. 50 ml cleared media or IFN-b standard in duplicate were

used for IFN-b ELISA by using human IFN-b ELISA kit (PBL

InterferonSource, NJ). For IL-6 production, the same media

samples were sent to The General Clinical Research Center

(GCRC) at Pennsylvania State University to measure IL-6 as

described in Lin et al [47].

Supporting Information

Figure S1 Interaction between PLK1 and Pcpi+. The cells were

transfected as in Fig. 1B. However, the cells were not metabolically

labeled and immunoprecipitated as in Fig. 1B. The immunopre-

cipitated peptides were resolved in SDS-PAGE and subjected to

immunoblotting using antibodies against P or PLK1.

Found at: doi:10.1371/journal.ppat.1000525.s001 (0.84 MB TIF)

Figure S2 Effect of PLK1 inhibitor GW843682 on PIV5 gene

expression. GW843682 (Sigma), a PLK1 inhibitor, was examined.

HeLa cells in 24-well plates were infected with rPIV5-RL at MOI

of 1 and incubated with GW843682 at 37uC for 16 to 20 hours.

Luciferase activities from the cells were measured as described in

Fig. 2A. The average of relative luciferase activity +/2 SEM is

shown.


Figure S3 Effects of P with S changed to A at position 157 (P-

S157A). (A). Interaction between P-S157A and PLK1. A plasmid

encoding Flag-PLK1 was transfected into cells with a plasmid

encoding P or P-S157A. The cells were metabolically labeled and

immunoprecipitated with anti-P or anti-Flag. (B). Effects of PLK1

overexpression on the mini-genome system using P-S157A.

Plasmids encoding pCAGGS-Flag-PLK1 at various concentra-

tions were transfected along with the mini-genome system. An

aliquot of cell lysate was used for immunoblotting to detect

expression levels of Flag-PLK1. The average of relative luciferase

activity +/2 SEM is shown.


Figure S4 Effects of mutating serine residues close to the serine

residue at position 308 of P on mini-genome gene expression. (A).

Comparison of the mini-genome activity of P to that of P with a S

to A change at position 304. (B). Comparison of the mini-genome

activity of P to that of P with a S to A change at position 313. The

average of relative luciferase activity +/2 SEM is shown.


Acknowledgments

We thank the members of Biao He’s laboratory for helpful discussion and

technical assistance. We are grateful to Dr. Michael Teng’s careful reading

of the manuscript.

Author Contributions

Conceived and designed the experiments: DS PL ZL BH. Performed the

experiments: DS PL ZL. Analyzed the data: DS PL BH. Wrote the paper:

DS BH.

References

1. Lamb RA, Kolakofsky D (2001) Paramyxoviridae: The viruses and their

replication. In: Knipe DM, Howley PM, eds. Fields Virology (Fourth Edition).

Philadelphia: Lippincott, Williams and Wilkins.

2. Lamb RA, Choppin PW (1977) The synthesis of Sendai virus polypeptides in

infected cells. III. Phosphorylation of polypeptides. Virology 81: 382–397.

3. Byrappa S, Pan YB, Gupta KC (1996) Sendai virus P protein is constitutively

phosphorylated at serine249: high phosphorylation potential of the P protein.

Virology 216: 228–234.

4. Hu CJ, Kato A, Bowman MC, Kiyotani K, Yoshida T, et al. (1999) Role of

primary constitutive phosphorylation of Sendai virus P and V proteins in viral

replication and pathogenesis. Virology 263: 195–208.

5. Hu C, Gupta KC (2000) Functional significance of alternate phosphorylation in

Sendai virus P protein. Virology 268: 517–532.

6. Lambert DM, Hambor J, Diebold M, Galinski B (1988) Kinetics of synthesis and

phosphorylation of respiratory syncytial virus polypeptides. J Gen Virol 69 (Pt 2):

313–323.

7. Navarro J, Lopez-Otin C, Villanueva N (1991) Location of phosphorylated

residues in human respiratory syncytial virus phosphoprotein. J Gen Virol 72 (Pt

6): 1455–1459.

8. Villanueva N, Navarro J, Mendez E, Garcia-Albert I (1994) Identification of a

protein kinase involved in the phosphorylation of the C-terminal region of

human respiratory syncytial virus P protein. J Gen Virol 75 (Pt 3): 555–

565.

9. Mazumder B, Adhikary G, Barik S (1994) Bacterial expression of human

respiratory syncytial viral phosphoprotein P and identification of Ser237 as the

site of phosphorylation by cellular casein kinase II. Virology 205: 93–103.



10. Sanchez-Seco MP, Navarro J, Martinez R, Villanueva N (1995) C-terminal

phosphorylation of human respiratory syncytial virus P protein occurs mainly atserine residue 232. J Gen Virol 76 (Pt 2): 425–430.

11. Lu B, Ma CH, Brazas R, Jin H (2002) The major phosphorylation sites of the

respiratory syncytial virus phosphoprotein are dispensable for virus replication invitro. J Virol 76: 10776–10784.

12. Asenjo A, Calvo E, Villanueva N (2006) Phosphorylation of human respiratorysyncytial virus P protein at threonine 108 controls its interaction with the M2-1

protein in the viral RNA polymerase complex. J Gen Virol 87: 3637–3642.

13. De BP, Gupta S, Banerjee AK (1995) Cellular protein kinase C isoform zetaregulates human parainfluenza virus type 3 replication. Proc Natl Acad Sc USA

92: 5204–5208.14. Huntley CC, De BP, Murray NR, Fields AP, Banerjee AK (1995) Human

parainfluenza virus type 3 phosphoprotein: identification of serine 333 as themajor site for PKC f phosphorylation. Virology 211: 561–567.

15. Chatziandreou N, Stock N, Young D, Andrejeva J, Hagmaier K, et al. (2004)

Relationships and host range of human, canine, simian and porcine isolates ofsimian virus 5 (parainfluenza virus 5). J Gen Virol 85: 3007–3016.

16. Evermann JF, Lincoln JD, McKiernan AJ (1980) Isolation of a paramyxovirusfrom the cerebrospinal fluid of a dog with posterior paresis. J Am Vet Med Assoc

177: 1132–1134.

17. Baumgartner WK, Krakowka S, Koestner A, Evermann J (1982) Acuteencephalitis and hydrocephalus in dogs caused by canine parainfluenza virus.

Vet Pathol 19: 79–92.18. Southern JA, Young DF, Heaney F, Baumgartner WK, Randall RE (1991)

Identification of an epitope on the P and V proteins of simian virus 5 thatdistinguishes between two isolates with different biological characteristics. J Gen

Virol 72: 1551–1557.

19. Chatziandreou N, Young D, Andrejeva J, Goodbourn S, Randall RE (2002)Differences in interferon sensitivity and biological properties of two related

isolates of simian virus 5: a model for virus persistence. Virology 293: 234–242.20. Wansley EK, Parks GD (2002) Naturally occurring substitutions in the p/v gene

convert the noncytopathic paramyxovirus simian virus 5 into a virus that induces

alpha/beta interferon synthesis and cell death. J Virol 76: 10109–10121.21. Timani KA, Sun D, Sun M, Keim C, Lin Y, et al. (2008) A single amino acid

residue change in the P protein of parainfluenza virus 5 elevates viral geneexpression. J Virol 82: 9123–9133.

22. Eckerdt F, Strebhardt K (2006) Polo-like kinase 1: target and regulator ofanaphase-promoting complex/cyclosome-dependent proteolysis. Cancer Res 66:

6895–6898.

23. Elia AE, Cantley LC, Yaffe MB (2003) Proteomic screen finds pSer/pThr-binding domain localizing Plk1 to mitotic substrates. Science 299: 1228–1231.

24. Sillje HH, Nigg EA (2003) Signal transduction. Capturing polo kinase. Science299: 1190–1191.

25. Elia AE, Rellos P, Haire LF, Chao JW, Ivins FJ, et al. (2003) The molecular basis

for phosphodependent substrate targeting and regulation of Plks by the Polo-boxdomain. Cell 115: 83–95.

26. Wu ZQ, Yang X, Weber G, Liu X (2008) Plk1 phosphorylation of TRF1 isessential for its binding to telomeres. J Biol Chem 283: 25503–25513.

27. Johnson EF, Stewart KD, Woods KW, Giranda VL, Luo Y (2007)Pharmacological and functional comparison of the polo-like kinase family:

insight into inhibitor and substrate specificity. Biochemistry 46: 9551–9563.

28. Sun M, Fuentes SM, Timani K, Sun D, Murphy C, et al. (2008) Akt plays acritical role in replication of nonsegmented negative-stranded RNA viruses.

J Virol 82: 105–114.

29. Datta A, Silverman L, Phipps AJ, Hiraragi H, Ratner L, et al. (2007) Human T-

lymphotropic virus type-1 p30 alters cell cycle G2 regulation of T lymphocytes to

enhance cell survival. Retrovirology 4: 49.

30. Bostik P, Dodd GL, Villinger F, Mayne AE, Ansari AA (2004) Dysregulation of

the polo-like kinase pathway in CD4+ T cells is characteristic of pathogenic

simian immunodeficiency virus infection. J Virol 78: 1464–1472.

31. Patel D, Incassati A, Wang N, McCance DJ (2004) Human papillomavirus type

16 E6 and E7 cause polyploidy in human keratinocytes and up-regulation of G2-

M-phase proteins. Cancer Res 64: 1299–1306.

32. Gallina A, Simoncini L, Garbelli S, Percivalle E, Pedrali-Noy G, et al. (1999)

Polo-like kinase 1 as a target for human cytomegalovirus pp65 lower matrix

protein. J Virol 73: 1468–1478.

33. Ludlow LE, Lo MK, Rodriguez JJ, Rota PA, Horvath CM (2008) Henipavirus V

Protein Association with Polo-Like Kinase Reveals Functional Overlap with

STAT1 Binding and Interferon Evasion. J Virol.

34. Lin GY, Lamb RA (2000) The paramyxovirus simian virus 5 V protein slows

progression of the cell cycle. J Virol 74: 9152–9166.

35. Niwa H, Yamamura K, Miyazaki J (1991) Efficient selection for high-expression

transfectants by a novel eukaryotic vector. Gene 108: 193–200.

36. Lin Y, Horvath F, Aligo JA, Wilson R, He B (2005) The role of simian virus 5 V

protein on viral RNA synthesis. Virology 338: 270–280.

37. He B, Lamb RA (1999) Effect of inserting paramyxovirus simian virus 5 gene

junctions at the HN/L gene junction: analysis of accumulation of mRNAs

transcribed from rescued viable viruses. J Virol 73: 6228–6234.

38. He B, Paterson RG, Ward CD, Lamb RA (1997) Recovery of infectious SV5

from cloned DNA and expression of a foreign gene. Virology 237: 249–260.

39. Li Z, Yu M, Zhang H, Magoffin DE, Jack PJ, et al. (2006) Beilong virus, a novel

paramyxovirus with the largest genome of non-segmented negative-stranded

RNA viruses. Virology 346: 219–228.

40. Baumann C, Korner R, Hofmann K, Nigg EA (2007) PICH, a centromere-

associated SNF2 family ATPase, is regulated by Plk1 and required for the

spindle checkpoint. Cell 128: 101–114.

41. Steegmaier M, Hoffmann M, Baum A, Lenart P, Petronczki M, et al. (2007) BI

2536, a potent and selective inhibitor of polo-like kinase 1, inhibits tumor growth

in vivo. Curr Biol 17: 316–322.

42. Lee KS, Yuan YLO, Kuriyama R, Erikson RL (1995) Plk Is an M-Phase-

Specific Protein-Kinase and Interacts with a Kinesin-Like Protein, Cho1/Mklp-

1. Molecular and Cellular Biology 15: 7143–7151.

43. Sun M, Fuentes SM, Timani K, Sun D, Murphy C, et al. (2008) Akt plays a

critical role in replication of nonsegmented negative-stranded RNA viruses.

Journal of Virology 82: 105–114.

44. Lansing TJ, McConnell RT, Duckett DR, Spehar GM, Knick VB, et al. (2007)

In vitro biological activity of a novel small-molecule inhibitor of polo-like kinase

1. Mol Cancer Ther 6: 450–459.

45. He B, Lin GY, Durbin JE, Durbin RK, Lamb RA (2001) The sh integral

membrane protein of the paramyxovirus simian virus 5 is required to block

apoptosis in mdbk cells. J Virol 75: 4068–4079.

46. Lin Y, Bright AC, Rothermel TA, He B (2003) Induction of apoptosis by

paramyxovirus simian virus 5 lacking a small hydrophobic gene. J Virol 77:

3371–3383.

47. Lin Y, Sun M, Fuentes SM, Keim CD, Rothermel T, et al. (2007) Inhibition of

interleukin-6 expression by the V protein of parainfluenza virus 5. Virology 368:

262–272.



JOURNAL OF VIROLOGY, Aug. 2009, p. 7948–7958 Vol. 83, No. 160022-538X/09/$08.00�0 doi:10.1128/JVI.00554-09Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Functional Characterization of Negri Bodies (NBs) in RabiesVirus-Infected Cells: Evidence that NBs Are Sites of Viral

Transcription and Replication�

Xavier Lahaye,1 Aurore Vidy,1† Carole Pomier,1 Linda Obiang,1 Francis Harper,2Yves Gaudin,1 and Danielle Blondel1*

CNRS, UMR2472, INRA, UMR1157, IFR 115, Virologie Moleculaire et Structurale, 91198, Gif sur Yvette, France,1 and CCNRS,FRE 2937, IFR 89, Laboratoire de Replication de l’ADN et Ultrastructure du Noyau, 94801 Villejuif, France2

Received 18 March 2009/Accepted 26 May 2009

Rabies virus infection induces the formation of cytoplasmic inclusion bodies that resemble Negri bodiesfound in the cytoplasm of some infected nerve cells. We have studied the morphogenesis and the role of theseNegri body-like structures (NBLs) during viral infection. The results indicate that these spherical structures(one or two per cell in the initial stage of infection), composed of the viral N and P proteins, grow during thevirus cycle before appearing as smaller structures at late stages of infection. We have shown that themicrotubule network is not necessary for the formation of these inclusion bodies but is involved in theirdynamics. In contrast, the actin network does not play any detectable role in these processes. These inclusionbodies contain Hsp70 and ubiquitinylated proteins, but they are not misfolded protein aggregates. NBLs, infact, appear to be functional structures involved in the viral life cycle. Specifically, using in situ fluorescenthybridization techniques, we show that all viral RNAs (genome, antigenome, and every mRNA) are locatedinside the inclusion bodies. Significantly, short-term RNA labeling in the presence of BrUTP strongly suggeststhat the NBLs are the sites where viral transcription and replication take place.

Viral infection can cause extensive cellular rearrangementsleading to the formation of cytoplasmic inclusions that concen-trate viral components. These inclusions are often consideredas side-products of the infectious process due to the passiveaccumulation of large quantities of proteins produced in excessduring infection. Nevertheless, in some cases, it has been dem-onstrated that these inclusions are in fact viral factories inwhich essential events of the viral cycle take place (30, 38).Indeed, in these compartments, the high concentrationof the viral components increases the efficiency of many steps ofthe viral infection (including transcription and/or replication ofthe genome and assembly of subviral particles) such that theyrepresent “viral factories.” Such factories have been identifiedfor a variety of unrelated viruses, including large DNA viruses(such as Poxviridae, Iridoviridae, and the closely related Africanswine fever virus), small DNA viruses (such as adenovirus andpoliomavirus), double-stranded RNA viruses (such as rotavi-ruses), and positive-stranded RNA viruses (such as togavirus)(39).

Similar inclusions called aggresomes are formed in responseto protein misfolding (7, 20). Proteins aggregates of misfoldedproteins are toxic to cells and are driven along microtubules toaggresomes for immobilization and subsequent degradation byproteasomes. The similarity between aggresomes and viral in-

clusions raises the possibility that viruses hijack the aggresomepathway to facilitate their own replication and assembly (14,39). Alternatively, aggresomes may be part of the innate cel-lular response that recognizes viral components as foreign ormisfolded and targets them for storage and degradation.

Rabies virus, the prototype of the lyssavirus genus that be-longs to the Rhabdoviridae family (Mononegavirales order),causes a fatal disease associated with intense viral replicationin the central nervous system. Its single-stranded negative-sense RNA genome (�12 kb), which encodes five viral pro-teins, is encapsidated by the nucleoprotein N (50 kDa) to formthe nucleocapsid that is associated with the RNA-dependentRNA polymerase L (220 kDa) and its cofactor the phospho-protein P (33 kDa). Inside the viral particle, this nucleocapsidhas a tightly coiled helical structure that is associated with thematrix protein M (22 kDa) and surrounded by a membranecontaining a unique glycoprotein G (62 kDa). The virus entersthe host cell through the endosomal transport pathway via alow-pH-induced membrane fusion process catalyzed by glyco-protein G (9). The nucleocapsid released into the cytoplasmserves as a template for transcription and replication processesthat are catalyzed by the L-P polymerase complex. Duringtranscription, a positive-stranded leader RNA and five cappedand polyadenylated mRNAs are synthesized. The replicationprocess yields nucleocapsids containing full-length antigenome-sense RNA, which in turn serve as templates for the synthesisof genome-sense RNA. During their synthesis, both the nas-cent antigenome and the genome are encapsidated by N pro-teins. The neosynthesized genome either serves as a templatefor secondary transcription or is assembled with M proteins toallow budding of the neosynthesized virion at a cellular mem-brane. Some electron microscopy studies suggest that the bud-

* Corresponding author. Mailing address: CNRS, UMR2472, INRA,UMR1157, IFR 115, Virologie Moleculaire et Structurale, 91198, Gifsur Yvette, France. Phone: 33 1 69 82 38 37. Fax: 33 1 69 82 43 08.E-mail: [email protected].

† Present address: Max von Pettenkofer Institute fur Virologie, Lud-wig Maximilians University, Pettenkoferstrasse 9a, 80336 Munich,Germany.

� Published ahead of print on 3 June 2009.

7948

at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

ding step occurs at the level of internal cellular compartments(27, 28).

Rabies virus infection induces the formation of cytoplasmicinclusion bodies (IBs) that are called Negri bodies (29). Sincethese structures are typical of rabies infection of the brain, theyhave a diagnostic value and have been used as definite histo-logical proof of such infection. These spherical structures hav-ing a diameter of a few micrometers (2 to 10 �m) are found inthe cytoplasm of some infected nerve cells. By electron micros-copy observations, the Negri bodies were found to be com-posed by a matrix of granular or filamentous material consist-ing of viral nucleoprotein. Viral particles were also seenbudding from this structure (21, 22, 24). Cytological stainingshave shown that they contain nucleic acids, suggesting that theymay be replication complexes. However, RNA labeling (with[3H]thymidine or [3H]uridine) failed to detect the IBs, arguingagainst this conclusion (23). Rabies virus RNA was detected byin situ hybridization in human brain, but no colocalization ofviral RNA and Negri bodies have been reported (17, 18).

In cultured cells from a neuronal or nonneuronal origin, IBsare also formed during viral infection. Based on their size andshape, it has been considered that these inclusions might beNegri body-like structures (NBLs).

In the present study, we have shown that the IBs formed ininfected cells do indeed have similar characteristics to those ofNegri bodies. Using this model, we have been able to study themechanisms involved in the formation of NBL structures ininfected cells and to investigate the role of these IBs duringviral infection, revealing that they may be factories where viralRNAs synthesis occurs.

MATERIALS AND METHODS

Cells and viruses. Human neuroblastoma SK cells, human glioblastoma as-trocytoma U373-MG cells and BSR cells cloned from BHK 21 (baby hamsterkidney) were grown in Dulbecco modified Eagle medium supplemented with10% fetal bovine serum.

The CVS strain of rabies virus was grown in BSR cells. Virus titers weredetermined by standard plaque assays into BSR cells.

Antibodies and drugs. The mouse polyclonal anti-P antibody has been de-scribed previously and the anti-P monoclonal antibodies (MAbs) 25C2 and 30F2have been previously described (35). The rabbit polyclonal anti-N antibody

conjugated to fluorescein isothiocyanate (FITC) was kindly provided by PierrePerrin (Institute Pasteur); anti-N MAb 62B5 was produced in mice immunizedwith the virus. Rabbit polyclonal anti-P antibody and mouse polyclonal anti-Mwere obtained by repeated injection of purified recombinant protein produced inEscherichia coli. Anti-G MAb 30AA5 has been previously described (8). Anti-HSp70 MAb (SPA-810) and anti-ubiquitin (SPA-203) were obtained fromStressgen. Anti-bromodeoxyuridine (anti-BrdU) MAb (clone B1C 9318) fromRoche and anti-� tubulin MAb (N356) from Amersham were used. Fluorescentsecondary antibodies were purchased from Molecular Probes, and anti-vimentin(V6630) and phalloidin (P1951) from Sigma. MitoTracker Red was obtainedfrom Invitrogen (M7513). Nocodazole (M1404), paclitaxel (T1402), cytochalasinD (catalog no. 30385), latrunculin A (L5163), and actinomycin D (ActD; A9415)were obtained from Sigma; jasplakinolide (420107) was purchased from Calbio-chem.

Plasmids. The Plasmids pCDM8-P encoding the P protein (CVS strain) andpCDM8-N encoding the N protein (CVS strain) have been described previously(3). The plasmid pP-RFP has been generated by cloning first the P gene in fusionwith the red fluorescent protein (RFP) gene in the plasmid pDsRedM (Clontech;gift of David Pasdeloup, MRC Virology Unit, Glasgow, United Kingdom), andthen the fusion product P-RFP was inserted into pcDNA-3 vector between NheIand XbaI restriction sites. The plasmid pP�143AQ147A-RFP was obtained fromthe previous construct pCDM8 encoding pP�143AQ147A (33) by using thestrategy described above. The dynamitin-green fluorescent protein (GFP)-ex-pressing construct (p50-GFP) was a generous gift from Richard Vallee (Colum-bia University, New York, NY) (5).

Cell infection and transfection. Monolayers of U373-MG cells or BSR cellswere grown on sterile glass coverslip in six-well plates (from 50 to 80% conflu-

FIG. 1. Rabies virus RNA species (genome, antigenome, andmRNAs encoding N, P, M, G, and L proteins) and location of senseand antisense primers (indicated by arrows) used in the FISH exper-iments. Abbreviations: Le, leader; Tr, trailer; An, poly(A).

TABLE 1. Sequences of the probes used in the FISH experimentsa

FISH probe Sequence (5�-3� biotinylated) First nucleotide positionb

Leader RNA GCTTTGCAATTGACGCTGTCTTTTTCTTCTTTGATCTGGTTG 9N mRNA CACCTGATTATTGACTTTGAACACAATCTTGTCGGCATCCAT 71P mRNA TATTTGCACTCCAACATTATCCAGGTACGATTGGAACAGGAG 1784M1 mRNA AGGAGAGGGCTTTTGAGTGTCCTCATCCCTACATTTTTTCAC 2517M2 mRNA CATAGTTGACATGCCCTCGAGTTCGAGTTGATACACCAGATC 2996G1 mRNA GCATCCTTCGTCCTCCACTACCAAATTGTTTGGGCAGCTGAG 3438G2 mRNA GGATCTCATTCCAAGTTCTGACTGACTTGTAGTGAGCATCGG 4348L1 mRNA TGGGTCAATAGGGTCATCATAGACCTCTCCAGGATCGAGCAT 5417L2 mRNA GTACTGATACAGAAACTCTAGGATTGCCTCCTCCTGTTGCTC 8820L3 mRNA CTGTTATACAGGTCGTGAAGTCTTGCGAAGCTAGGGACGTCA 11470N genome ATGGATGCCGACAAGATTGTGTTCAAAGTCAATAATCAGGTG 71P genome CTCCTGTTCCAATCGTACCTGGATAATGTTGGAGTCCAAATA 1784Antigenome GAGAAAAACAATCTCACACCAGAGGTTCGGATTCAAGATCTT 11862Poly-A d(T)x42M mRNA VSV CCTAATTTCTTAGATTTCTTAACCTTTCCCCTTCAGACCGAG 2271GADPH mRNA TGGCCAAGGTCATCCATGACAACTTTGGCATTGTGGAAGG

a The name of the probe corresponds to the viral RNA detected (the probe N mRNA is used to detect the N mRNA).b Nucleotides are numbered according to their distance from the 5� end of rabies virus and VSV virus antigenome (positive strand).

VOL. 83, 2009 FUNCTIONAL CHARACTERIZATION OF RABIES VIRUS NEGRI BODIES 7949

at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

ence) and infected at different multiplicity of infection (MOI) of rabies virus(CVS strain) and at different times postinfection (p.i.).

U373-MG cells or BSR cells were grown on sterile glass coverslip in six-wellplates (from 50 to 80% confluence) and were transfected with 2.5 �g or 5 �g ofplasmid DNA by the calcium phosphate coprecipitation procedure. For N and Pcoexpression, the T7 vaccinia virus system was used as described previously (3).

Treatments of cells with drugs. Cells were kept in Dulbecco modified Eaglemedium containing 2 �M nocodazole, 1.25 nM paclitaxel, 2.5 �M cytochalasinD, 1 �M latrunculin A, or 0.5 �M jasplakinolide for 1 h before virus inoculationand during infection. Since all drugs were reconstituted in dimethyl sulfoxide(DMSO), untreated cells were maintained in medium containing the same con-centration of DMSO. Cells were fixed and processed for microscopy.

Immunofluorescence staining and confocal microscopy. Cells monolayersgrown on glass coverslip were fixed 20 min with 4% (wt/vol) paraformaldehyde inphosphate-buffered saline (PBS) and permeabilized for 5 min in 0.1% Triton X-100in PBS. Cells were then prepared for double-immunofluorescence staining andanalyzed by confocal microscopy. The intracellular distribution of viral antigen P, N,M, and G were analyzed by using the specific antibody at the dilution of 1/500 to1/1,000, followed by incubation with the corresponding immunoglobulin G (IgG)antibody conjugated to Alexa 568 or Alexa 488 at the dilution of 1/1,000. The cellswere mounted onto glass slides by using Immu-Mount (Shandon) containing DAPI(4�,6�-diamidino-2-phenylindole) to stain nuclei.

Confocal laser microscopy was performed on a Leica SP2 microscope (�63oil-immersion objective lens) using UV excitation at 351 nm (DAPI) and bluelaser excitation at 488 nm (Alexa 488) and green laser excitation at 545 nm(Alexa 568) in sequential recording mode.

Sellers’ staining. Infected cells grown on coverslips were stained with a mix-ture of saturated solution of basic fuschin and methylene blue dissolved inmethanol as described by Sellers in 1927 (40). Cells were fixed and permeabilizedwith methanol at �20°C, dipped into the staining solution for several seconds,and then washed with water. Cells were observed under a light microscope.

Quantification of Negri body-like structures. BSR cells were infected withCVS at different MOIs (0.5, 1, 3, 5, and 10) in the presence or not of nocodazole(2 �M). Cells were fixed and permeabilized at different times p.i. (4 to 16 h). Theviral N protein was stained by a rabbit polyclonal anti-N conjugated to FITC.Quantification of IBs was performed by counting the viral inclusions (containingthe N protein) present in infected cells. An average of 400 cells from threeindependent experiments were counted at each time p.i. and for each MOI. Theresults are expressed as a percentage of uninfected cells and cell containing one,two, three, four, and more IBs.

Poisson distribution predicts that the relative frequency (p) of one cell to beinfected by n virus at an MOI of is p (n) (e� � n)/n!.

FISH. Fluorescence in situ hybridization (FISH) was performed by using 3�biotinylated oligonucleotides (biotin 569.61, obtained from Eurogentec) (42nucleotides) complementary to specific sequences of viral RNA (Table 1 andFig. 1). According to the amount of viral RNA expected to be present duringrabies virus infection, the number of probes mixed during hybridization wasdifferent: the NmRNA and PmRNA were detected by using one probe, theMmRNA and GmRNA were detected by using a mixture of two probes (M1and M2 or G1 and G2), the LmRNA was detected by using three probes (L1,L2, and L3), and the genome was detected by using a mixture of two probes.The probe used to detect the antigenome was designed in order to avoid thedetection of the mRNA, whereas the probe used to detect the mRNA hy-bridizes also to the antigenome.

Infected cells grown on coverslips were fixed with paraformaldehyde andpermeabilized as described above. After treatment with 200 U of DNase I

FIG. 2. Characterization of the viral IBs. (A) BSR cells were in-fected with rabies virus (CVS strain) at an MOI of 3 for 16 h andstained by the Sellers’ staining protocol as described in Materials andMethods. Viral IBs are indicated pointed by arrows (left panel). Ahigher magnification of an area is also presented (right panel). The

scale bar corresponds to 12 �m. (B) BSR cells were infected as de-scribed above. After fixation and permeabilization, cells were pro-cessed for double-immunofluorescence staining with a rabbit poly-clonal anti-P antibody (P) and the MAb 62B5 anti-N antibody (N), amouse polyclonal anti-M antibody (M), or the MAb 30AA5 anti-Gantibody (G). In all cases, DAPI (blue) was used to stain the nuclei(merge). Colocalization is apparent as yellow coloration in the mergedpanel. The scale bars correspond to 12 �m. (C) Infected BSR cells asdescribed above were treated for immunogold labeling. P protein wasdetected by using two MAbs, 25C2 and 30F2; N protein was detectedby using MAb 64B6; and M and G proteins were detected by usingmouse polyclonal anti-M antibody and by anti-G MAb 30AA5, respec-tively. The scale bars correspond to 0.5 �m.

7950 LAHAYE ET AL. J. VIROL.

at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

(RNase-free D7201; Sigma)/ml, cells were prehybridized in hybridization buffer(50% formamide, 10% dextran sulfate, 4� SSC [1� SSC is 0.15 M NaCl plus0.015 M sodium citrate]) for 30 min at 37°C before being incubated with 10 �l ofhybridization mix containing 10 ng of biotinylated probe, 10 �g of herring testisDNA, and 8.5 �l of hybridization buffer for 2 min at 95°C in presence or not of15 U of RNase A (catalog no. 10109142001; Roche)/ml. Coverslips were then putin a humidified hybridization chamber at 37°C overnight and washed three timesat 42°C for 5 min with 50% formamide–2� SSC and then three times at 42°Cwith 0.5� SSC. Cells were then processed for immunofluorescence staining. TheN protein was stained with a mouse MAb anti-N (62B5) antibody, followed byincubation with Alexa 488-goat anti-mouse IgG. Viral biotinylated RNA weredetected by incubation with streptavidin conjugated to cyanin 3 (Cy3; catalog no.01 61 600 81; Jackson Immunotech).

BrUTP labeling of viral RNA. BrUTP incorporation of newly synthezised viralRNA was performed as described by Georgi et al. (11). Briefly, the infected cells(16 h p.i.) were treated with 15 �g of ActD/ml for 1 h and then transfected withBrUTP (B 7166; Sigma) at a final concentration of 10 mM by using Lipo-fectamine 2000 (Invitrogen) and maintained in the presence of ActD for 20 min.The cells were fixed and permeabilized as described above and processed forimmunofluorescence. P protein was detected by using the rabbit polyclonalanti-P antibody, followed by incubation with Alexa 568-conjugated goat anti-rabbit IgG. The BrUTP-labeled RNAs were stained by using mouse MAb toBrdU at a dilution of 1/50, followed by the addition of goat anti-mouse Alexa488-conjugated antibody.

Electron microscopy. For structural analysis, infected BSR cells were fixed for1 h at room temperature with 2% glutaraldehyde (Sigma) in 0.1 M cacodylatebuffer (pH 7.4; Sigma). Cells were washed with 0.1 M sucrose (Sigma) in 0.1 Mcacodylate buffer (pH 7.4). Cells were then postfixed for 1 h at room temperaturewith 1% aqueous osmium tetroxide (OsO4) and 1.5% potassium ferrocyanide(Sigma) in 0.1 M cacodylate buffer (pH 7.4) and stained with 2% uranyl acetatein water. The cells were then dehydrated in increasing concentrations of ethanoland embedded by Epon with 2,4,6-tris(dimethylaminomethyl) phenol (DMP30;Delta Microscopies). Polymerization was carried out for 48 h at 60°C. Ultrathin

sections of Epon-embedded material were collected on copper palladium grids(200 mesh) and stored until use. These sections were stained with lead citrate(Delta Microscopies). Sections were examined with a Zeiss EM902 electronmicroscope operated at 80 kV, and images were acquired with a charge-coupleddevice camera (Megaview III) and analyzed with ITEM software (Eloïse, France;MIMA2 platform [INRA-CRJ]).

For immunogold detection of rabies virus protein, cells were fixed for 1 h at4°C with 4% formaldehyde (Merck) in 0.1 M phosphate buffer (pH 7.3). Duringfixation, cells were scraped and centrifuged. Pellets were dehydrated in increas-ing concentrations of methanol and embedded in Lowicryl K4M (ChemischeWerke Lowi) at a low temperature according to the method of Roth et al. (36).Polymerization was carried out for 5 days at �30°C under long-wavelength UVlight (Phillips fluorescent tubes [TL 6W]). Ultrathin sections of Lowicryl-embed-ded material were collected on carbon-coated gold grids (200 mesh) and storeduntil use. For immunogold labeling, grids bearing Lowcryl thin sections of in-fected cells were first placed for 2 min over drops of bovine serum albumin (5%PBS). The grids were then floated for 2 h, at room temperature, on drops ofaffinity purified mouse monoclonal or polyclonal anti-rabies virus proteins anti-bodies (diluted 1/20 in PBS) or over normal mouse serum as control (diluted 1/20in PBS). After 15-min washes with 3 drops of PBS, the grids were incubated for30 min over drops of a 1/30 dilution in PBS of goat anti-mouse IgG conjugatedto gold particles (10 nm in diameter; Biocell Research Laboratories). After5-min passages on 3 drops of PBS, the grids were rapidly rinsed in jet-distilledwater, air dried, and stained for 10 min with 5% aqueous uranyl acetate. Gridswere observed with a Philips 400 transmission electron microscope, at 80 kV atmagnifications of �8,000 to 22,000.

RESULTS

Characterization of the IBs found in rabies virus-infectedcells. Rabies virus infection induces the formation of round IBsin the cytoplasm of neuronal and nonneuronal cells. We ob-

FIG. 3. Evolution of the viral IBs during the viral cycle. (A) BSR cells were infected by CVS at an MOI of 3 for different times p.i. as indicatedand stained with rabbit polyclonal anti-N conjugated to FITC antibody and DAPI. The scale bars correspond to 12 �m. (B and C) The numbersof IBs per cell were quantified at different times p.i. for an MOI of 3 (B) or an MOI of 10 (C). A histogram represents the average values fromthree independent experiments by counting 400 cells per experiment. The results are indicated as the percentages of cells that were uninfected orthat contained one, two, three, four, and more IBs and are depicted by six variously shaded columns from left to right, respectively, for each timepoint. Error bars indicate the standard deviations.


at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

served IBs in infected BSR cells (Fig. 2). Consistent with theknown properties of Negri bodies, these inclusions could bestained with the Seller’s solution containing the dye, basicfuschin (40) (Fig. 2A).

Unsurprisingly, confocal and electron microscopy analysesrevealed that the IBs are full of N and P proteins (Fig. 2B andC). It has also been shown that the polymerase L is present inthe IBs (6). In contrast and as expected, the M and G proteinswere excluded from these structures, as indicated by the factthat IBs were neither labeled by a polyclonal serum against theM protein nor by an MAb directed against the G protein (Fig.2B and C). Thus, the staining characteristics, the size (a fewmicrometers), and the shape, as well as the viral protein com-position of the IBs formed in infected BSR cells, indicated thatthese structures share the defined features as Negri bodies andcan thus be considered NBLs.

Morphogenesis of the NBLs during viral infection. Little isknown about the formation and evolution of Negri bodiesduring viral infection. Our model system permits us to performsuch characterization. This is the first point that we have ad-dressed.

BSR cells or human neuroblastoma SK cells were infected atdifferent MOIs (0.5 to 10) and analyzed at different times p.i.by immunofluorescence using anti-N or anti-P antibodies. At

different times p.i., IBs in the cytoplasm of cells infected atdifferent MOIs were counted. At an early stage of infection (4h p.i.), infected cells presented one or two viral inclusions thatthen grew in size (10 to 12 h p.i.) (Fig. 3A). These values wereobserved regardless of the MOI used (between 3 and 10).Smaller IB structures appeared at later stage of infection (16 to20 h p.i.) (Fig. 3A).

The number of NBLs per cell was independent of the MOIand consequently was independent of the number of incomingvirus per cell (Fig. 3B and C) that was expected to follow aPoisson distribution as mentioned in the Materials and Meth-ods. These results suggest that there are a limited number ofsites per cell that allow the formation of IBs.

The morphogenesis and structural evolution of NBLs werealso analyzed in infected BSR cells by electron microscopy(Fig. 4). NBLs are spherical structures (a few micrometers insize) that often contain a cavity. Although initially devoid ofmembranes (4 to 12 h p.i., Fig. 4A), NBLs progressively be-came associated with a double membrane (16 h p.i., Fig. 4Band C). The cytoplasmic granular surface of the membranesuggests that it might be derived from rough endoplasmic re-ticulum (Fig. 4C). Finally, NBLs were completely surroundedby these double membranes (16 to 24 h p.i.). This membraneorganization is reminiscent of the organization of the double

FIG. 4. Morphology and evolution of viral IBs. Ultrastructural aspect of BSR cells infected for 12 h (A), 16 h (B and C), and 24 h (D and E)p.i. as indicated. Double fixation was performed with glutaraldehyde and osmium tetraoxide. Epon embedding and uranium-lead staining. IBsdisplay a granular dense structure with a cavity (indicated by arrows in panels A, B, C, and E). Note the recruitment of membranes around IBsat 16 h p.i. (B and C) and the presence of virus particles close to the viral inclusion at 24 h p.i. (indicated by arrow in panel D). Nu, nucleus; Cy,cytoplasm. Bars correspond to 0.5 �m. Higher magnification (�13) of two membrane areas reveals a double-membrane with a granular surfaceare shown for panel C.


at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

membrane around the cell nucleus. After 24 h p.i., the shape ofthe bodies appeared to be altered, and viral particles were alsoseen budding from some of these structures (Fig. 4D and E),most probably inside the lumen of the compartments thatsurround the bodies.

Comparison of the NBLs and aggresomes. Aggresomes aremisfolded protein aggregates that appear when the protea-

some and chaperones can no longer deal with the aggregatedproteins. They are driven on microtubules to the microtubuleorganizing center (MTOC), they are surrounded by vimentincages to isolate and reduce the toxicity of misfolded proteins,and they recruit mitochondria that may provide the ATP re-quired by chaperones and proteasomes for protein folding anddegradation (7, 20). For many viral families, it has been pro-posed that the aggresome pathway is hijacked in order toconcentrate viral components and to facilitate viral replicationand assembly (14, 39). Therefore, we investigated whetherNBLs share the characteristics of aggresomes described above.We observed that NBLs do not colocalize with the MTOC, asrevealed by anti-�-tubulin staining (Fig. 5A). Costaining withanti-vimentin antibodies did not show any rearrangement ofvimentin around the cytoplasmic inclusions at any stage ofinfection (Fig. 5A). Furthermore, we did not observe recruit-ment of mitochondria in close proximity to the NBLs (Fig. 5A).However, Hsp70 chaperone and ubiquitinylated proteins accu-mulate inside the IBs, as revealed by an anti-Hsp70 and anti-poly-ubiquitin staining, respectively (Fig. 5B). These resultsindicate that the NBLs do not share all of the features ofaggresomes, although they contain Hsp70 and ubiquitinylatedproteins.

Role of the cytoskeleton in the morphogenesis of NBLs. Wenext investigated the mechanism by which the different pro-teins gather and are recruited in NBL structures by analyzingthe role of the cytoskeleton in the morphogenesis and evolu-tion of the IBs.

We studied the role of the microtubule network by using twodifferent specific inhibitors of MT dynamics (nocodazole andpaclitaxel). BSR or U373-MG infected cells (MOI 3 andMOI 10) were treated with nocodazole, which results in thedisruption of microtubules, for 1 h before virus inoculation andduring infection. In the presence of nocodazole, the “early”(large, one or two per cell) NBLs formed efficiently withoutany apparent delay (Fig. 6). These results indicate that micro-tubules are not needed for the IB formation.

In striking contrast, the formation of “late” (small) IBs wasaffected by nocodazole treatment since at later stages of infec-tion (16 to 20 h p.i.), we observed NBLs that were much largerthan those in the absence of the drug. Furthermore, at theselate stages, no smaller structures were observed in nocodazole-treated cells (Fig. 6). This was in contrast to untreated cellswhere small IBs (more than 15 per cell) were detected. Thus,the drug inhibited the formation of additional IBs and/or theappearance of smaller structures.

We did not observe any effect of paclitaxel, a drug thatstabilizes microtubules, on the formation and evolution of IBs(data not shown). This suggests that the mechanisms involvedin these processes do not require dynamic microtubules.

As P protein has been shown to interact with LC8, the lightchain of the molecular motor dynein (19, 34), we checkedwhether this interaction plays a role in the targeting of themajor P component to the viral inclusions. When cells wereinfected and transfected with the plasmid encoding the P pro-tein mutant unable to interact with LC8 (P�LC8-RFP) (33),RFP fluorescent viral inclusions were formed as in cells in-fected and transfected with the plasmid encoding the wild-typeP-RFP protein (Fig. 7A). This indicates that LC8 is not re-quired in the recruitment of P inside the IBs. This is consistent

FIG. 5. Viral IBs exhibit some features of aggresomes but are dis-tinct structures. (A) BSR cells were infected for 16 h and costainedwith a rabbit polyclonal anti-P antibody (P) and either mouse anti-�tubulin MAb (Tub), mouse anti-vimentin MAb (Vim), or the Mito-Tracker Red (Mito), followed by incubation with Alexa 488-goat anti-rabbit IgG and Alexa 568-goat anti-mouse IgG. Note that IBs are notlocalized at the MTOC (indicated by arrows) and that no vimentincage or no recruitment of mitochondria (as would be expected foraggresomes) are observed around the IBs. (B) IBs contain Hsp70 andubiquitinylated proteins. Infected BSR cells were immunostained witha rabbit polyclonal FITC-conjugated anti-N antibody (N) and a mouseMAb anti-ubiquitin antibody (Ub) or a mouse MAb anti-Hsp70 anti-body (Hsp70), followed by incubation with Alexa 568-goat anti-mouseIgG. The scale bars correspond to 12 �m.


at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

with the fact that overexpression of the inhibitor of dynein-mediated transport, dynamitin (dynamitin-GFP) during infec-tion did not result in the inhibition of the formation of viralinclusions (Fig. 7B).

The role of actin was also investigated in BSR cells or inU373-MG cells by using inhibitors resulting in depolymeriza-tion of actin (cytochalasin D and latrunculin A) or resulting instabilization of actin (jasplakinolide). None of the inhibitorssignificantly affected the formation and evolution of NBLs,indicating that actin filaments are not required in these pro-cesses.

NBLs are potential sites of viral transcription and replica-tion. We then analyzed whether NBLs are functional struc-tures where some essential steps of the virus cycle take place.Since the viral N, P, and L proteins involved in viral transcrip-tion and replication accumulate in IBs, we investigated thepresence of viral RNA in these structures by using FISH. Wetherefore designed biotinylated oligonucleotide probes to de-tect specifically viral mRNAs, antigenomic RNA, and genomicRNA (Table 1 and Fig. 1).

After 16 h of infection, BSR cells were simultaneously pre-pared for FISH to detect the viral RNA and immunostainedwith anti-N antibody to visualize the Negri bodies.

All viral mRNAs encoding the five viral proteins (N, P, M,

G, and L) were detected inside the IBs with no viral mRNAreadily detected elsewhere (i.e., nucleus/cytoplasm) outside ofIBs (Fig. 8A). The relative quantities of the viral RNAs weredetermined (Fig. 8D). Although the FISH method does notallow a precise quantification, it clearly indicated that the NmRNA was more abundant than the L mRNA according to theknown gradient of decreasing mRNAs concentration that re-flects the gene order (N, P, M, G, and L) (1). These hybrid-izations were specific since no positive signal was observed inmock-infected cells (data not shown) or in infected cellstreated with a probe detecting the mRNA of the M protein ofvesicular stomatitis virus (VSV) (Fig. 8C). A positive controlcarried out in infected cells with a poly(dT) probe or a probeagainst the housekeeping GADPH mRNA gave rise to a signaldiffusely distributed in the nucleus and the cytoplasm as ex-pected (Fig. 8C). RNase A treatment of infected cells prior toFISH with poly(dT) probe prevented the hybridization signalin the nucleus and the cytoplasm but left intact the signal in theIBs (Fig. 8C). These results indicate that the viral mRNAaccumulate inside the IBs and are protected against RNasedigestion, most probably by the cage formed by the N and Pproteins. Indeed, the N protein formed an apparent ringlikestructure (cage) around the RNA (Fig. 2B and Fig. 8A).

By using specific probes, the viral genomic RNA and anti-

FIG. 6. Effect of nocodazole on formation and evolution of IBs during infection. (A) BSR cells were infected and treated with nocodazole (2�M; NCZ) or mock treated (DMSO). At different times p.i. as indicated, the cells were costained with rabbit polyclonal anti-N conjugated to FITCantibody and with mouse MAb anti-� tubulin antibody, followed by incubation with Alexa 568-goat anti-mouse IgG. The scale bars correspondto 12 �m. (B and C) The numbers of IBs per cell were quantified in nocodazole-treated cells at different times p.i. as indicated for an MOI of 3(B) or an MOI of 10 (C). A histogram represents the average values from three independent experiments by counting 400 cells per experiment.The results are expressed as the percentages of uninfected cells or cells containing one, two, three, four, and more IBs and are depicted by sixvariously shaded columns from left to right, respectively, for each time point. Error bars indicate the standard deviation.


at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

genomic RNA were also detected exclusively in the IBs (Fig.8B). The relative quantification of the viral RNAs indicatedthat the amount of genomic RNA is higher than the amount ofantigenomic RNA (Fig. 8D).

The experiments described above demonstrated that NBLsare structures where viral transcription and replication prod-ucts accumulate. In order to establish whether the viral RNAsare synthesized inside IBs or synthesized outside IBs and thentranslocated into these structures, we performed short-termRNA labeling in the presence of BrUTP and ActD to inhibitcellular transcription. Cells were infected for 16 h before treat-ment without or with ActD for 1 h and transfected with BrUTPfor 20 min. Cells were then examined by double-immunofluo-rescence staining using anti-BrdU and anti-P antibodies (Fig.9). Without ActD, cellular RNA was mainly detected in thenucleus, as expected (Fig. 9A, upper panel). In the presence ofthe drug, newly synthesized RNA were detected inside the viralinclusions (Fig. 9A, lower panel). Control experiments wereperformed in which rabies virus N and P proteins were coex-pressed (with the vaccinia virus T7 system) to mimic NBLs, aspreviously shown (3), and provide nonviral IBs. The fact thatno BrUTP signal was detected in these nonviral IBs in thepresence of the drug indicated that the BrUTP signals were

dependent on the incorporation of BrUTP by the rabies virusRNA polymerase (Fig. 9B). These results provide evidencethat NBs are functional structures where viral transcriptionand replication take place.

DISCUSSION

The present study characterizes the morphogenesis andfunctions of the IBs that are found in the cytoplasm of cellsinfected by rabies virus. We show that these IBs have stainingcharacteristics that are similar to those of the Negri bodiesfound in certain nerve cells infected by rabies virus. Thesestructures are not canonical aggresomes containing misfoldedproteins aggregates en route to degradation by the proteasome(consistent with the observations of Menager et al. [26]) butare functional structures with a central role in viral infection.

The detection of viral RNA in IBs indicates either that viralRNAs are synthesized within IBs or are synthesized elsewherebefore import and accumulation within IBs. Our findings indi-cate that the former is the case and that Negri bodies are mostprobably specialized sites of viral transcription and replication.Significantly, the labeling period used in the BrUTP experimentsis short compared to the duration of viral mRNA synthesis. In-deed, it has been reported that due to a quite slow elongation rateof transcription (3.1 to 4.3 nucleotides), it takes about 10 min forthe VSV N mRNA and 1 to 2 h for the LmRNA to be synthesized(16). This is also consistent with the elongation rates recentlyreported for another Mononegavirales, the measles virus (threenucleotides) (32). Thus, the newly BrUTP-labeled RNAs are al-most certainly synthesized within NBLs (during the 20-min label-ing period), as indicated by the BrUTP labeling experiment pre-viously used for VSV (4).

This would be consistent with the fact that proteins L, P,and N involved in viral RNA synthesis are accumulated inthese structures. Interestingly, viral genomes, antigenomes,and mRNAs synthesized inside the IBs appear to be sur-rounded by a cage formed by N and P proteins that mayprotect them from degradation. We did not investigatewhether NBLs are also sites of viral protein synthesis. Whetherthe viral proteins L, N, and P are translated inside or at theperiphery of the viral inclusions or anywhere else, and in thislatter case how these proteins are recruited in such IBs, re-mains an open question. Cellular proteins are also detected inthese structures. Although the exact composition of the NBLsremains unknown, the Toll-like receptor 3 has been shown tobe an important compound of the NBLs (26). In addition, theheat shock protein Hsp70 and ubiquitinylated proteins, whichare also found associated with aggresomes, were accumulatedinside these structures. This enrichment might be correlatedwith an intense production of viral proteins (particularly N andP) in these viral factories. However, in addition to continuouslysurveying the folding status of proteins as part of its qualitycontrol function, Hsp70 is also involved in many cellular pro-cesses such as cell cycle, apoptosis, or innate immunity (25).Similarly, ubiquitinylation of proteins is also involved in manycellular functions other than degradation, such as signalizationevents. Since viruses need to interfere with cellular processesto create a favorable environment for their multiplication,Hsp70 and ubiquitin ligases are frequently diverted and re-cruited by viruses (2, 25). In the case of rabies virus infection,

FIG. 7. Formation of IBs does not require dynein-mediated trans-port. (A) BSR cells were transfected with a plasmid encoding theP-RFP or a P-RFP mutant that does not interact with LC8 (P�LC8-RFP) and then infected for 24 h. Cells were stained with the MAb62B5 anti-N antibody (N), followed by incubation with Alexa 488-goatanti-mouse IgG and DAPI (merge). The scale bars correspond to 12�m. (B) BSR cells were transfected with a plasmid encoding dyna-mitin-GFP and then infected for 24 h. Cells were stained with a rabbitpolyclonal anti-P antibody (P), followed by incubation with Alexa568-goat anti-rabbit IgG and DAPI (merge). Colocalization is appar-ent as yellow coloration in the merged panel. The scale bars corre-spond to 12 �m.


at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

Hsp70 might play a proviral role, as previously indicated by thepresence of Hsp70 in purified rabies virus (37). In addition,NEDD4, an E3 ubiquitin ligase, is recruited by rabies virusmatrix protein via a PPXY motif and is involved in the buddingprocess (13, 15).

Electron-microscopy studies indicate that viral inclusionsinitially devoid of membranes recruit cellular membranes, sug-gesting that the budding step could occur inside intracellularcompartment derived from the rough endoplasmic reticulumand wrapped around IBs as previously shown (24).

The fact that IBs may support viral transcription and repli-cation leads us to consider that such structures may formaround incoming genomes. Our results indicate that early afterinfection, infected cells contain one or two viral factories. Thislimited initial number of formation sites is independent of theMOI and consequently is independent of the number of in-coming viruses per cell. The mechanism involved in this re-striction is not known, but it might occur at the cell entry tolimit the number of incoming viral particles and/or inside thecytoplasm to reduce the number of functional transcriptionand replication sites. It is also possible that a viral factory arisesfrom many incoming genomes. The restriction would then bedue to the requirement of specific cellular structures or or-ganelles (present in a limited number in the cell) for the IBs toform.

We did not observe any preferential intracellular localiza-tion of the viral inclusions in the cells such as the MTOC,where aggresomes are typically located, as well as some otherviral factories (31). The lack of a link between MTOC andrabies viral factories is consistent with the fact that microtubulenetworks are not involved in the morphogenesis of NBLs thatare formed efficiently and without any delay in the presenceof an inhibitor of microtubule polymerization. Furthermore,

FIG. 8. Viral RNAs accumulate in viral IBs. BSR cells were in-fected (MOI 3) for 16 h p.i. (A and B) FISH analyses were per-formed with biotinylated oligonucleotides (described in Materials andMethods), followed by incubation with streptavidin conjugated to Cy3to detect NmRNA, PmRNA, MmRNA, GmRNA, and LmRNA asindicated (A) and to detect viral genomic and antigenomic RNA asindicated (B). Note that the probe used to detect the mRNA alsoallows the detection of the antigenome. In contrast, the probe used todetect the antigenome was designed in order to avoid the detection ofthe mRNA. Note that for the antigenomes, the photomultiplier tube(PMT) has been increased in order to clearly visualize the fluorescencesignals resulting in an increase of the background signal. No positivesignal was observed in mock-infected cells (indicated by an arrow).(C) FISH was also performed to detect the housekeeping GADPHmRNA or cellular mRNA (Poly-A) after treatment with RNase(�RNase) or without RNase (–RNase). No signal was observed with acontrol probe complementary to the MmRNA of VSV. In all cases, theN protein (N) was stained with the mouse MAb anti-N (62B5) anti-body, followed by incubation with Alexa 488-goat anti-mouse IgG (A,B, and C). The scale bars correspond to 12 �m. (D) Relative quanti-fication of the viral RNA accumulated inside the Negri bodies. FISHwas performed as described above. Quantification of Cy3 fluorescentintensity (arbitrary units [ua]) was performed with the same PMT (thePMT adjusted with the N mRNA probe) on a Leica SP2 confocalmicroscope. The surface area and the Cy3 fluorescence intensity ofeach Negri body was calculated by using Leica confocal software. Twoindependent experiments were performed by counting an average of300 IBs. The results were expressed as the log10 of Cy3’s intensity per�m2 of Negri body and per probe (ua/�m2).


at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

overexpression of dynamitin that blocks dynein/dynactin-me-diated retrograde transport along microtubules has no effecton the morphogenesis of viral inclusions. All of these dataindicate that the mechanism to gather the different viral andcellular components in viral inclusions does not require anactive cellular transport along the microtubules.

Although microtubule depolymerization does not inhibit the

formation of Negri bodies, it prevents the appearance ofsmaller IBs that otherwise form late in infection. Further ex-periments are required to establish whether these small inclu-sions are derived from the initial IBs or may be newly formedby infection at a later stage by new viral progeny or just bydelayed formation. In contrast, the actin network does not playa prominent role in these processes.

It is possible that NBLs might be the result of a cytoplasmiccompartmentalization due to a cellular defense mechanisminvolving components of the aggresome pathway, such as hostfactors involved in stress response, the storage or degradationof proteins to target virus components for storage, and degra-dation, as previously suggested for other viruses. These struc-tures may be hijacked by the rabies virus to concentrate hostand viral components in one place to facilitate viral replication,to immobilize and inactivate proteins that would otherwiseinhibit infection, and also to mask rabies virus presence fromthe cellular components of the innate immune system. This isconsistent with the findings of Menager et al. (26), who sug-gested that the sequestration of Toll-like receptor 3 in theNegri body could be a strategy whereby the virus prevents theantiviral or apoptotic effect of this cellular protein. Cytoplas-mic inclusions have been reported for filoviruses (10) and someparamyxoviruses (12), suggesting that virus-induced formationof intracellular compartments might be general among virusesof the Mononegavirales order that contains numerous humanand animal pathogens. It is probable that the mechanisms andthe machineries hijacked by rabies virus are similarly divertedby these other viruses.

ACKNOWLEDGMENTS

We are grateful to Greg Moseley (from Monash University, Clayton,Victoria, Australia) for helpful discussions and careful reading of themanuscript. We thank Manos Mavrakis for helpful discussions. Wethank David Pasdeloup for providing the plasmid pDsRedM and fortechnical advice for the FISH. We acknowledge Richard Vallee for thegift of the plasmid encoding dynamitin-GFP. We acknowledgeChristine Longin for helpful advice and assistance with the electronmicroscopy (Unite UR1196 Genomique et Physiologie de la Lac-tation, INRA, Plateau de Microscopie Electronique MIMA2, Jouy-en-Josas Cedex, France). Confocal microscopy was performed onthe Plate-forme Imagerie et Biologie Cellulaire of the CNRS cam-pus supported by the Institut Federatif de Recherche 87, La Planteet Son Environnement, and the program ASTRE of the ConseilGeneral de l’Essonne.

We acknowledge support from the CNRS and INRA.

REFERENCES

1. Ball, L. A., and C. N. White. 1976. Order of transcription of genes ofvesicular stomatitis virus. Proc. Natl. Acad. Sci. USA 73:442–446.

2. Chen, M., and D. Gerlier. 2006. Viral hijacking of cellular ubiquitinationpathways as an anti-innate immunity strategy. Viral Immunol. 19:349–362.

3. Chenik, M., K. Chebli, Y. Gaudin, and D. Blondel. 1994. In vivo interactionof rabies virus phosphoprotein (P) and nucleoprotein (N): existence of twoN-binding sites on P protein. J. Gen. Virol. 75(Pt. 11):2889–2896.

4. Das, S. C., D. Nayak, Y. Zhou, and A. K. Pattnaik. 2006. Visualization ofintracellular transport of vesicular stomatitis virus nucleocapsids in livingcells. J. Virol. 80:6368–6377.

5. Dohner, K., A. Wolfstein, U. Prank, C. Echeverri, D. Dujardin, R. Vallee,and B. Sodeik. 2002. Function of dynein and dynactin in herpes simplex viruscapsid transport. Mol. Biol. Cell 13:2795–2809.

6. Finke, S., K. Brzozka, and K. K. Conzelmann. 2004. Tracking fluorescence-labeled rabies virus: enhanced green fluorescent protein-tagged phospho-protein P supports virus gene expression and formation of infectious parti-cles. J. Virol. 78:12333–12343.

7. Garcia-Mata, R., Z. Bebok, E. J. Sorscher, and E. S. Sztul. 1999. Charac-terization and dynamics of aggresome formation by a cytosolic GFP-chimera.J. Cell Biol. 146:1239–1254.

FIG. 9. Newly synthesized viral RNAs are present inside the viralIBs. (A) BSR cells were infected at an MOI of 3 for 16 h p.i., trans-fected with BrUTP for 20 min in the presence (�ActD) or absence(–ActD) of ActD. P protein (as a marker of the Negri bodies) wasdetected by using the rabbit polyclonal anti-P antibody, followed byincubation with Alexa 568-goat anti-rabbit IgG (P). Newly synthesizedRNAs were detected by using mouse MAb anti-BrdU antibody(BrUTP) and goat anti-mouse Alexa 488. Colocalization of the viralRNA (vRNA) with the IBs is shown in the merged panel in thepresence of the drug (�ActD). In cells untreated with ActD, cellularmRNA is produced in vast excess over vRNA. Thus, for these images,the PMT gain of the confocal laser scanning microscope is lower, sothe signal for vRNA is lower than for that shown in the presence of thedrug, and specific colocalization is not readily observed. The scale barscorrespond to 12 �m. (B) BrUTP signals are specific to RNA producedby rabies virus. BSR cells were infected with the VTF7-3 recombinantvirus and cotransfected with plasmids expressing N and P proteins.Cells were then transfected with BrUTP for 20 min in the presence(�ActD) or absence (–ActD) of ActD. P protein and RNA weredetected as in panel A. Note that in the absence of the drug, no BrUTPsignal is detectable in the IBs formed by N and P proteins, but someRNA signals are detected in the cytoplasmic inclusions, most probablydue to vaccinia virus infection (–ActD). In the presence of the drug, noRNA labeling is observed (�ActD).


at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

8. Gaudin, Y. 1997. Folding of rabies virus glycoprotein: epitope acquisitionand interaction with endoplasmic reticulum chaperones. J. Virol. 71:3742–3750.

9. Gaudin, Y. 2000. Rabies virus-induced membrane fusion pathway. J. CellBiol. 150:601–612.

10. Geisbert, T. W., and P. B. Jahrling. 1995. Differentiation of filoviruses byelectron microscopy. Virus Res. 39:129–150.

11. Georgi, A., C. Mottola-Hartshorn, A. Warner, B. Fields, and L. B. Chen.1990. Detection of individual fluorescently labeled reovirions in living cells.Proc. Natl. Acad. Sci. USA 87:6579–6583.

12. Goldsmith, C. S., T. Whistler, P. E. Rollin, T. G. Ksiazek, P. A. Rota, W. J.Bellini, P. Daszak, K. T. Wong, W. J. Shieh, and S. R. Zaki. 2003. Elucida-tion of Nipah virus morphogenesis and replication using ultrastructural andmolecular approaches. Virus Res. 92:89–98.

13. Harty, R. N., M. E. Brown, J. P. McGettigan, G. Wang, H. R. Jayakar, J. M.Huibregtse, M. A. Whitt, and M. J. Schnell. 2001. Rhabdoviruses and thecellular ubiquitin-proteasome system: a budding interaction. J. Virol. 75:10623–10629.

14. Heath, C. M., M. Windsor, and T. Wileman. 2001. Aggresomes resemblesites specialized for virus assembly. J. Cell Biol. 153:449–455.

15. Irie, T., J. M. Licata, J. P. McGettigan, M. J. Schnell, and R. N. Harty. 2004.Budding of PPxY-containing rhabdoviruses is not dependent on host pro-teins TGS101 and VPS4A. J. Virol. 78:2657–2665.

16. Iverson, L. E., and J. K. Rose. 1981. Localized attenuation and discontinuoussynthesis during vesicular stomatitis virus transcription. Cell 23:477–484.

17. Jackson, A. C., D. L. Reimer, and W. H. Wunner. 1989. Detection of rabiesvirus RNA in the central nervous system of experimentally infected miceusing in situ hybridization with RNA probes. J. Virol. Methods 25:1–11.

18. Jackson, A. C., and W. H. Wunner. 1991. Detection of rabies virus genomicRNA and mRNA in mouse and human brains by using in situ hybridization.J. Virol. 65:2839–2844.

19. Jacob, Y., H. Badrane, P. E. Ceccaldi, and N. Tordo. 2000. Cytoplasmicdynein LC8 interacts with lyssavirus phosphoprotein. J. Virol. 74:10217–10222.

20. Johnston, J. A., C. L. Ward, and R. R. Kopito. 1998. Aggresomes: a cellularresponse to misfolded proteins. J. Cell Biol. 143:1883–1898.

21. Manghani, D. K., D. K. Dastur, A. N. Nanavaty, and R. Patel. 1986. Pleo-morphism of fine structure of rabies virus in human and experimental brain.J. Neurol. Sci. 75:181–193.

22. Matsumoto, S. 1962. Electron microscopy of nerve cells infected with streetrabies virus. Virology 17:198–202.

23. Matsumoto, S. 1970. Rabies virus. Adv. Virus Res. 16:257–301.24. Matsumoto, S., L. G. Schneider, A. Kawai, and T. Yonezawa. 1974. Further

studies on the replication of rabies and rabies-like viruses in organizedcultures of mammalian neural tissues. J. Virol. 14:981–996.

25. Mayer, M. P. 2005. Recruitment of Hsp70 chaperones: a crucial part of viralsurvival strategies. Rev. Physiol. Biochem. Pharmacol. 153:1–46.

26. Menager, P., P. Roux, F. Megret, J. P. Bourgeois, A. M. Le Sourd, A.Danckaert, M. Lafage, C. Prehaud, and M. Lafon. 2009. Toll-like receptor 3(TLR3) plays a major role in the formation of rabies virus Negri bodies.PLoS Pathog. 5:e1000315.

27. Murphy, F. A., S. P. Bauer, A. K. Harrison, and W. C. Winn, Jr. 1973.Comparative pathogenesis of rabies and rabies-like viruses. Viral infectionand transit from inoculation site to the central nervous system. Lab. Investig.28:361–376.

28. Murphy, F. A., A. K. Harrison, W. C. Winn, and S. P. Bauer. 1973. Com-parative pathogenesis of rabies and rabies-like viruses: infection of the cen-tral nervous system and centrifugal spread of virus to peripheral tissues. Lab.Investig. 29:1–16.

29. Negri, A. 1903. Contributo allo studio dell’eziologia della rabia. Bol. Soc.Med. Chir. Pavia 1903:88–114.

30. Novoa, R. R., G. Calderita, R. Arranz, J. Fontana, H. Granzow, and C. Risco.2005. Virus factories: associations of cell organelles for viral replication andmorphogenesis. Biol. Cell 97:147–172.

31. Nozawa, N., Y. Yamauchi, K. Ohtsuka, Y. Kawaguchi, and Y. Nishiyama.2004. Formation of aggresome-like structures in herpes simplex virus type2-infected cells and a potential role in virus assembly. Exp. Cell Res. 299:486–497.

32. Plumet, S., W. P. Duprex, and D. Gerlier. 2005. Dynamics of viral RNAsynthesis during measles virus infection. J. Virol. 79:6900–6908.

33. Poisson, N., E. Real, Y. Gaudin, M. C. Vaney, S. King, Y. Jacob, N. Tordo,and D. Blondel. 2001. Molecular basis for the interaction between rabiesvirus phosphoprotein P and the dynein light chain LC8: dissociation ofdynein-binding properties and transcriptional functionality of P. J. Gen.Virol. 82:2691–2696.

34. Raux, H., A. Flamand, and D. Blondel. 2000. Interaction of the rabies virusP protein with the LC8 dynein light chain. J. Virol. 74:10212–10216.

35. Raux, H., F. Iseni, F. Lafay, and D. Blondel. 1997. Mapping of monoclonalantibody epitopes of the rabies virus P protein. J. Gen. Virol. 78:119–124.

36. Roth, J., D. J. Taatjes, and K. T. Tokuyasu. 1990. Contrasting of LowicrylK4M thin sections. Histochemistry 95:123–136.

37. Sagara, J., and A. Kawai. 1992. Identification of heat shock protein 70 in therabies virion. Virology 190:845–848.

38. Wileman, T. 2006. Aggresomes and autophagy generate sites for virus rep-lication. Science 312:875–878.

39. Wileman, T. 2007. Aggresomes and pericentriolar sites of virus assembly:cellular defense or viral design? Annu. Rev. Microbiol. 61:149–167.

40. Young, C. C., and T. F. Sellers. 1927. Laboratory: a new method for stainingNegri bodies of rabies. Am. J. Public Health 17:1080–1081.


at NO

RT

HW

ES

TE

R U

NIV

ER

SIT

Y on N

ovember 2, 2009

jvi.asm.org

Dow

nloaded from

http://jvi.asm.org

REUL Is a Novel E3 Ubiquitin Ligase and Stimulator ofRetinoic-Acid-Inducible Gene-IDong Gao, Yong-Kang Yang, Rui-Peng Wang, Xiang Zhou, Fei-Ci Diao, Min-Dian Li, Zhong-He Zhai,

Zheng-Fan Jiang, Dan-Ying Chen*

Key Laboratory of Cell Proliferation and Differentiation (Ministry of Education), College of Life Sciences, Peking University, Beijing, China

Abstract

RIG-I and MDA5 are cytoplasmic sensors that recognize different species of viral RNAs, leads to activation of thetranscription factors IRF3 and NF-kB, which collaborate to induce type I interferons. In this study, we identified REUL, a RING-finger protein, as a specific RIG-I-interacting protein. REUL was associated with RIG-I, but not MDA5, through its PRY andSPRY domains. Overexpression of REUL potently potentiated RIG-I-, but not MDA5-mediated downstream signalling andantiviral activity. In contrast, the RING domain deletion mutant of REUL suppressed Sendai virus (SV)-induced, but notcytoplasmic polyI:C-induced activation of IFN-b promoter. Knockdown of endogenous REUL by RNAi inhibited SV-triggeredIFN-b expression, and also increased VSV replication. Full-length RIG-I, but not the CARD domain deletion mutant of RIG-I,underwent ubiquitination induced by REUL. The Lys 154, 164, and 172 residues of the RIG-I CARD domain were critical forefficient REUL-mediated ubiquitination, as well as the ability of RIG-I to induce activation of IFN-b promoter. These findingssuggest that REUL is an E3 ubiquitin ligase of RIG-I and specifically stimulates RIG-I-mediated innate antiviral activity.

Citation: Gao D, Yang Y-K, Wang R-P, Zhou X, Diao F-C, et al. (2009) REUL Is a Novel E3 Ubiquitin Ligase and Stimulator of Retinoic-Acid-Inducible Gene-I. PLoSONE 4(6): e5760. doi:10.1371/journal.pone.0005760

Editor: Jerome Nigou, Institut de Pharmacologie et de Biologie Structurale CNRS 205, France

Received January 3, 2009; Accepted May 6, 2009; Published June 1, 2009

Copyright: ! 2009 Gao et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricteduse, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This work was supported by the National Natural Science Foundation of China (30421004 and 30871288), the Foundation for Authors of NationalExcellent Doctoral Dissertations of PR China (200533) and the China 973 Program (2006CB504301). The funders had no role in study design, data collection andanalysis, decision to publish, or preparation of the manuscript.

Competing Interests: The authors have declared that no competing interests exist.

* E-mail: [email protected]

Introduction

The innate immune system plays critical roles in recognizingviral infections, and triggers signaling cascades that induce anti-viral mediators such as type I interferons (IFNs) and pro-inflammatory cytokines [1]. Type I IFNs induce the expressionof a set of IFN-stimulated genes that inhibit viral replication ininfected cells as well as in neighboring uninfected cells.Transcriptional activation of the promoters of type I IFN genesrequires the coordinated activation of multiple transcriptionfactors and their cooperative assembly into transcriptionalenhancer complexes in vivo. The enhancer of the IFN-b genecontains a kB site recognized by nuclear factor kB (NF-kB), a sitefor ATF-2/c-Jun, and two IFN-stimulated response elements(ISREs) recognized by phosphorylated interferon regulatory factor(IRF)-3 and/or IRF-7 [2].The innate immune system has evolved at least two distinct

mechanisms for the recognition of viral RNAs [1]. One ismediated by membrane-bound Toll-like receptors, which areimportant for the production of type I IFNs in plasmacytoiddendritic cells (pDCs). The second mechanism involves cytosolicreceptors for RNAs, retinoic-acid-inducible gene-I (RIG-I) andmelanoma differentiation associated protein-5 (MDA5), whichplay essential roles in the recognition of RNA viruses in variouscells other than pDCs. Gene-knockout studies indicate that RIG-Iand MDA5 are required for responding to distinct species of RNAviruses. RIG-I responds to in vitro-transcribed dsRNA, vesicularstomatitis virus (VSV), Newcastle disease virus (NDV), and

influenza virus in mice. In contrast, MDA5 recognizes poly(I:C)and is essential for the antiviral response to the picornavirusencephalomyocarditis virus [3,4]. Furthermore, RIG-I, but notMDA5, recognizes single-strand RNA bearing 59 phosphate [5,6].Both RIG-I and MDA5 belong to the DExD/H box RNA

helicase family, and contain two CARD modules at their Nterminus and a DexD/H-box helicase domain at their C terminus.The helicase domains of RIG-I and MDA5 serve as intracellularviral RNA receptors, whereas the CARD modules are responsiblefor transmitting signals to the downstream CARD-containingadaptor VISA/MAVS/IPS-1/Cardif, which in turn activatesTAK1-IKKa and TBK1/IKKb kinases, leading to activation ofNF-kB and IRF-3 and induction of type I IFNs [7–12].As with other cytokine systems, several mechanisms are thought

to underlie the positive and negative regulation of RIG-I signaling.It has recently become clear that RIG-I is regulated by ubiquitinconjugation mediated by a RING finger family protein, RNF125,an E3-ubiquitin ligase that specifies its proteosomal degradation[13]. Expression of RNF125 increases (Lys)-48-linked polyubiqui-tin and destabilization of RIG-I, while the exact locus forRNF125-mediated RIG-I ubiquitination is not known. Recentstudies have shown that polyubiquitination of signaling proteinsthrough lysine (Lys)-63-linked polyubiquitin chains plays animportant role in the activation of NF-kB [14], while it was alsoshown that RIG-I undergoes (Lys)-63-linked ubiquitination at itsN-terminal 2CARD. The Lys 172 residue of RIG-I is critical forefficient ubiquitination and for VISA binding, as well as the abilityof RIG-I to induce antiviral signal transduction. And the K63-

PLoS ONE | www.plosone.org 1 June 2009 | Volume 4 | Issue 6 | e5760

linked ubiquitin is delivered by tripartite motif 25 (TRIM25) E3ligase [15].In the present study, we identified a RING-finger protein,

REUL, as a novel RIG-I E3 ubiquitin ligase. REUL wasspecifically associated with RIG-I through its PRY and SPRYdomains, and this interaction effectively resulted in a markedincrease of RIG-I downstream signaling activity. Furthermore, theLys 154, 164, and 172 residues of the RIG-I CARD domain weredetermined to be critical for efficient REUL-mediated ubiquitina-tion and for the ability of RIG-I to induce antiviral signaltransduction. These findings suggest that REUL is an E3 ubiquitinligase of RIG-I and specifically stimulates RIG-I-mediated innateantiviral signaling.

Results

Identification of REULIt is well established that RIG-I is an essential cytosolic sensor

for the innate immune response to certain viruses. However, theregulatory mechanism of RIG-I-mediated signaling has not beenadequately characterized. To identify potential proteins thatinteract with RIG-I and regulate its signaling, we performed yeasttwo-hybrid screens of a human fetal kidney cDNA library usingfull-length RIG-I as bait. From 56106 independent screenedclones, we identified 17 b-galactosidase-positive clones, amongwhich 5 clones encoded the C-terminal regions of RNF135 (Aa125–432, GenBank accession number NM_032322.3, and alsoidentified as Riplet in recent publication [16]). This gene is locatedin a chromosomal region known to be frequently deleted inpatients with neurofibromatosis [17–19]. On the basis of itsfunctions described below, we designated this protein as REUL(for RIG-I E3 ubiquitin ligase). Since the REUL clone obtainedfrom the yeast two-hybrid screening was not full-length, weamplified full-length REUL cDNA from the human fetal kidneycDNA library using PCR. Human REUL contains 432 amino acidresidues and shares 57% sequence identity at the amino acid levelwith its mouse ortholog (Fig. 1A). Structural analysis with severalprograms indicated that REUL contains an N terminus RING-finger domain, and a C terminus PRY and SPRY domain(Fig. 1A). The middle region of REUL has no detectable similarityto any other proteins.To confirm whether REUL is expressed in mammalian cells at

the protein level, we raised a mouse antiserum against humanREUL. Western blot analysis showed that REUL was expressed asa 48 kDa protein in all examined human cell lines, includingHEK293, HT1080, Hela, U937, K562, and HUVEC cells. Thesize of the endogenous REUL protein was slightly smaller thanthat of FLAG-tagged REUL overexpressed in HEK293 cells(Fig. 1B). These data suggest that REUL is expressed at the proteinlevel.We next determined the cellular localization of REUL.

Immunofluorescence experiments showed that overexpressedand endogenous REUL were localized in the cytoplasm ofHEK293T and Hela cells (Fig. 1C).

Interaction between RIG-I and REULBecause REUL interacted with RIG-I in the yeast two-hybrid

system, we further determined whether full-length REULinteracted with RIG-I in mammalian cells. Coimmunoprecipita-tion experiments in HEK293T cells principally indicated thatFlag-tagged RIG-I associated with HA-tagged REUL in co-transfection experiments (Fig. 2A). This interaction was specificbecause in the same experiment REUL did not interact with

MDA5, which is structurally and functionally related to RIG-I.REUL also did not interact with adaptor protein VISA (Fig. 2A).REUL was subsequently shown to colocalize with RIG-I in co-

transfected HEK293T and Hela cells (Fig. 2B). Double immuno-fluorescent staining showed that overexpressed REUL had asimilar distribution pattern and overlapped with RIG-I, but notVISA.We next determined whether REUL interacts with RIG-I in

untransfected cells and the effects of viral infection on theinteraction. Endogenous coimmunoprecipitation experimentsindicated that RIG-I did not interact with REUL withoutstimulation, but was associated with REUL after SV infection.As a control, REUL did not interact with RIG-I under transfectedpolyI:C stimulation (Fig. 2C). These results suggest that REUL isassociated with RIG-I in a viral-infection-dependent manner.REUL contains an N-terminal RING-finger domain, and C-

terminal PRY and SPRY domains (Fig. 2D). To determine whichdomain facilitated interaction with RIG-I, we made REULdeletion mutants. Coimmunoprecipitation assays showed that themutants carrying PRY and SPRY domains (91–432 and 215–432)bound to RIG-I as effectively as full-length REUL, whereas theRING-finger domain mutant (1–90), the middle region mutant(91–214) and the mutant carrying the RING-finger and the middleregion (1–214) did not interact with RIG-I (Fig. 2D). These resultssuggest that the PRY and SPRY domains of REUL are requiredfor interaction with RIG-I.RIG-I contains two CARD modules at the N terminus and a

DexD/H-box helicase domain at the C terminus. Similarly, wemade RIG-I deletion mutants and determined which domain isrequired for interaction with REUL. Domain-mapping experi-ments indicated that the mutant carrying the N terminal CARDdomain (1–218) and the mutant carrying the DexD/H-boxhelicase domain (219–925) interacted with REUL individually(Fig. 2E). Since REUL has similar domain structure withTRIM25, we also detected the interactions between TRIM25and RIG-I truncations for comparison. Coimmunoprecipitationexperiments demonstrated that, in the same way, TRIM25interacted with the helicase domain of RIG-I as efficiently asCARD domain of RIG-I. As control, in the same conditions theRIG-I full length and truncated RIG-I did not interact withTRIM5-a, which has a similar structure to TRIM25 and REUL(Fig. 2E). These results suggest that there are multiple binding siteson RIG-I which mediate the interaction with REUL andTRIM25.

REUL potentiates RIG-I signalingBecause REUL is specifically associated with RIG-I, we

examined whether REUL is involved in regulation of RIG-I-mediated signaling. In reporter assays, overexpression of REULalone had no apparent effect on activation of ISRE, NF-kB andIFN-b promoter, while in cotransfection experiments, RIG-I-mediated induction of ISRE, NF-kB and IFN-b promoter activityconsiderably increased with REUL expression in a dose-dependent manner (Fig. 3A, B, C). In these assays, REUL didnot increase MDA5-mediated activation of ISRE, NF-kB andIFN-b promoter (Fig. 3A, B, C).Dimerization and translocation are hallmarks of IRF3 activa-

tion. Consistently, overexpression of REUL potentiated RIG-I-but not MDA5-mediated dimerization of IRF-3 (Fig. 3D). Doubleimmunofluorescence staining showed that overexpression ofREUL caused translocation of IRF3 into the nucleus inHEK293 cells stably expressing RIG-I as efficiently as overex-pression of VISA, but this did not happen in HEK293 cells stablyexpressing MDA5 (Fig. 3E).

Novel E3/Stimulator of RIG-I


We further tested whether REUL increased RIG-I-mediatedantiviral and pro-inflammatory gene expression. In RT-PCRexperiments, we found that cotransfection of REUL potentlyincreased RIG-I-, but not MDA5-induced expression of endoge-nous antiviral and pro-inflammatory genes, including IFN-b,regulated on activation, normal T-cell expressed and secreted

(RANTES), interferon-stimulated gene 15 (ISG15) and TNFa(Fig. 3F). These data suggest that REUL specifically increasesRIG-I- but not MDA5-mediated signaling.Notably, we found that expression of the RING domain

deletion mutant REULDRING (Aa 91–432), which was sufficientto bind to RIG-I, markedly suppressed RIG-I-, but not MDA5-

Figure 1. Sequence and expression of REUL. (A) Alignment of human and mouse REUL amino acid sequences. The RING, PRY and SPRY domainsare indicated by green, blue and red lines individually. (B) Expression of REUL in mammalian cells. Lysates of the indicated cells were analyzed byWestern blot with a mouse polyclonal anti-REUL antibody. (C) REUL localization in HEK293 and Hela cells. Upper panels: HEK293 cells transfected withan expression plasmid for HA-tagged REUL, immunofluorescent staining was performed with anti-HA (aHA) or mouse IgG; untransfected HEK293 cellsstained with preimmune serum (preserum) or anti-REUL serum (aREUL); lower panels: transfected or untransfected Hela cells stained as in the upperpanels. The experiments were repeated twice, and similar results were obtained.doi:10.1371/journal.pone.0005760.g001



Figure 2. REUL associates with RIG-I. (A) REUL interacts with RIG-I but not MDA5 and VISA. HEK293 cells (16106) were transfected with HA-REULalone or together with Flag-RIG-I, Flag-MDA5 and Flag-VISA plasmids (5 mg each). Cell lysates were immunoprecipitated (IP) with anti-Flag (aF), anti-HA (aHA) or mouse IgG antibody (IgG). The immunoprecipitates were analyzed by Western blot (IB) with anti-HA or anti-Flag antibody. Whole-celllysates (WCL) were analyzed by Western blots with anti-HA and anti-Flag to determine the expression of REUL, RIG-I, MDA5 and VISA. (B) REUL iscolocalized with RIG-I but not VISA in HEK293T and Hela cells. Upper panels: HEK293T cells were transfected with HA-REUL, Flag-RIG-I or Flag-VISA.Immunofluorescent staining was performed with anti-HA (red) and anti-Flag (green); lower panels: Hela cells were transfected and stained as in the



mediated ISRE, NF-kB and IFN-b promoter in a dose-dependentmanner (Fig. 3G, H, I). These data suggest that the RING domainof REUL is required for its function to potentiate RIG-I-mediatedsignaling.

Role of REUL in RIG-I antiviral activityPreviously, it has been demonstrated that RIG-I responds to

infection by SV, VSV, NDV, and influenza virus, whereas MDA5is involved in IFN signaling triggered by poly (I:C) transfected intothe cytoplasm [3,19].We further determined whether REULplayed a role in the cellular antiviral response mediated by RIG-I.In reporter gene assays, overexpression of REULDRING (Aa 91–432) markedly inhibited SV-induced activation of ISRE, NF-kBand IFN-b promoter in a dose-dependent manner (Fig. 4A, B, C).In contrast, overexpression of REULDRING had no apparenteffect on IFN-b promoter activation triggered by cytoplasmic poly(I:C) in HEK293T cells (Fig. 4D). These results confirmed thatREUL is positively and specifically involved in RIG-I-, but notMDA5-mediated antiviral activity.We next determined whether endogenous REUL is required for

RIG-I-mediated antiviral activity. We utilized three REUL-RNAiplasmids targeting different sites of human REUL mRNA.Transient transfection and Western blot analysis showed thattwo of these RNAi plasmids (2# and 3#) markedly inhibited theexpression of transfected REUL in HEK293T cells, whereas the1# RNAi plasmid had little effect on REUL expression (Fig. 4E).In reporter gene assays, knockdown of REUL by 2# and 3#suppressed SV-induced activation of ISRE, NF-kB and IFN-bpromoter as efficiently as VISA RNAi plasmid, whereas the 1#had no apparent effect on SV-induced signaling (Fig. 4F, G, H).Consistent with this, we found that transfection of REUL RNAiplasmids 2# and 3# potently inhibited SV-induced expression ofendogenous IFN-b and RANTES in RT-PCR experiments(Fig. 4I).We further determined whether REUL plays a role in the

cellular antiviral response. On infection by newcastle disease virus-enhanced green fluorescent protein (NDV-eGFP), REUL expres-sion significantly suppressed NDV-eGFP replication in HEK293cells, whereas expression of REULDRING detectably increasedNDV-eGFP replication. We noted that combined expression ofRIG-I and REUL had the strongest effect on suppression ofNDV–eGFP replication, and this effection was comparable tooverexpression of VISA (Fig. 4J). Using plaque assays, we alsofound that, cotransfection of REUL efficiently enhanced theinhibitory effect on vesicular stomatitis virus (VSV) replicationmediated by overexpression of RIG-I (Fig. 4K), whereasknockdown of REUL had the opposite effect, increasing viralreplication as potently as VISA RNAi (Fig. 4L).Collectively, these data suggest that REUL plays a critical role

in the efficient cellular antiviral response mediated by RIG-I.

REUL is an E3 ubiquitin ligase of RIG-IPreviously, it has been reported that RIG-I undergoes robust

ubiquitination at its N-terminal 2CARD, and RIG-I ubiquitina-

tion is induced by the TRIM family member TRIM25 inmammalian cells [15]. As the RING domain of REUL is requiredfor potentiation of RIG-I-mediated signaling, we predicted thatREUL is a novel E3 ubiquitin ligase of RIG-I. To test the role ofREUL in RIG-I ubiquitination, RIG-I was co-expressed withwild-type REUL or the E3 ligase-defective REULDRING (Aa 91–432) mutant together with ubiquitin. We found that RIG-I wasextensively ubiquitinated and REUL expression markedly in-creased the ubiquitination levels of RIG-I. In contrast, REULDR-ING efficiently decreased RIG-I ubiquitination levels (Fig. 5A). Toexclude that Trim25 is ultimately affecting RIG-I ubiquitination,we repeated this experiment using Trim25 RNAi and found thatin TRIM25 knock-down cells, REUL facilitated the ubiquitinationof RIG-I efficiently (Fig. 5B). Finally, in vitro ubiquitination assayconfirmed that REUL (E3), together with E1 and UbcH5b (E2),effectively delivered the ubiquitin to RIG-I (Fig. 5C). These resultsdemonstrated that REUL directly affects RIG-I ubiquitination,and the effect of REUL is independent of TRIM25.Furthermore, we detected that SV infection treatment resulted

in the markedly increased ubiquitination of endogenously RIG-I.Overexpression of REUL accelerated SV induced ubiquitinationof endogenously RIG-I, and REULDRING potently inhibited thiseffect (Fig. 5D). These results confirmed REUL effections onubiquitination of endogenous RIG-I.To demonstrated which domain of RIG-I was ubiquitinated by

REUL, HEK293T cells were co-transfected with Flag-taggedREUL, HA-tagged full-length RIG-I, or a RIG-I mutant RIG-IDCARD (Aa 219–925), which the 2CARD had been deleted,together with Flag-tagged ubiquitin. REUL expression markedlyincreased the ubiquitination levels of full-length RIG-I, but notRIG-IDCARD (Fig. 5E). These data suggested that REULubiquitinates the CARD domain of RIG-I.

The Lys 154, 164, and 172 residues of the RIG-I CARDdomain are critical for efficient REUL-mediatedubiquitination and signaling activity of RIG-IPreviously, it has been reported that the Lys 172 residue in the

RIG-I CARD domain is critical for TRIM25-mediated ubiquiti-nation and the ability of RIG-I to induce antiviral signaltransduction [15]. To further dissect the possible ubiquitinationsites of RIG-I that are critically involved in antiviral signaling, weindividually replaced the nineteen lysine residues in the RIG-ICARD domain with arginine. In reporter gene assays, we foundthat compared with wild-type RIG-I, the K154R, K164R andK172R caused marked loss of the ability to induce activation ofIFN-b promoter. In contrast, the other KR mutations: K18, 45,48, 59, 95, 96, 99, 108, 115, 146, 169, 177, 181, 190, 193 and203R had little or no effect on IFN-b promoter activation (Fig. 6A).We next tested the functions of mutants K154R, K164R,

K172R individually and in various combinations. Reporter assayresults showed that the double mutants K164/154R, K172/154R,K172/164R, and the triple mutant K154R/K164R/K172Rcaused near-complete loss of the ability to induce activation of

upper panels. The experiments were repeated twice, and similar results were obtained. (C) Endogenous interaction of REUL and RIG-I. HEK293T(36106) cells were treated with SV, transfected with polyI:C (100 mg) or left untreated for 4 h before lysate. Cell lysates were immunoprecipitated (IP)and Western blot (IB) by indicated antibodies. (D) Identification of domains of REUL mediating the interaction with RIG-I. HEK293T cells (16106) weretransfected with expression plasmids for HA-RIG-I alone or together with Flag-REUL or its mutants (5 mg each). Cell lysates were immunoprecipitatedwith indicated antibodies. The immunoprecipitates were analyzed by Western blots with anti-HA or anti-Flag antibody. Whole-cell lysates wereanalyzed by Western blots with anti-Flag and anti-HA to determine the expression levels of transfected plasmids. Upper panel: schematicrepresentation of REUL. (E) Identification of domains of RIG-I mediating the interaction with REUL. HEK293T cells (16106) were transfected with F-REUL (left panel), F-TRIM25 (middle panel), F-TRIM5-a (right panel) alone or together with HA-RIG-I or its mutants. Immunoprecipitation and Westernblot analysis were performed with indicated antibodies.doi:10.1371/journal.pone.0005760.g002



Figure 3. REUL potentiates RIG-I signaling. (A), (B), (C) REUL potentiates RIG-I-, but not MDA5-mediated activation of ISRE (A), NF-kB (B) and IFN-b (C) promoter in a dose-dependent manner. HEK293T cells (16105) were transfected with the indicated reporter plasmid (50 ng), pRL-TK Renillaluciferase plasmid (50 ng), expression plasmid for RIG-I or MDA5 (200 ng) and the indicated amounts of an expression plasmid for REUL or emptyplasmids. Luciferase assays were performed 24 h after transfection. (D) REUL potentiates RIG-I-, but not MDA5-mediated IRF3 dimerization. HEK293cells (26105) were transfected with indicated plasmids. 24 h after transfection, cell lysates were separated by native PAGE and analyzed with IRF3antibody. The experiments were repeated three times with similar results. (E) REUL potentiates RIG-I-, but not MDA5-mediated IRF3 translocation.



IFN-b promoter. As control, the K18R and double mutantK190R/K193R induced IFN-b promoter activity as strongly aswild-type RIG-I (Fig. 6B). These data suggested that Lys 154, 164and 172 are the critical sites for RIG-I signaling activity.We also determined whether Lys 154, 164, and 172 are the key

sites for REUL-mediated ubiquitination and activity of RIG-I.Reporter assay results showed that overexpression of REULpotently potentiated the ability of wild-type RIG-I to activate IFN-b promoter. But the K154R, K164R and K172R mutationssignificantly reduced the activation level, and REUL almostcompletely failed to induce activation of RIG signaling whencotransfected with the double mutants K164/154R, K172/154R,or K172/164R, or the triple mutant K154R/K164R/K172R(Fig. 6B). In contrast, co-expression with REUL, K18R andK190R/193R induced IFN-b activity as strongly as the wild-type.These data indicated that Lys 154, 164, and 172 are the key sitesfor the REUL-mediated signaling activity of RIG-I.To test the role of Lys 154, 164, and 172 in RIG-I

ubiquitination mediated by REUL, wild type RIG-I andmutations were co-expressed with or without REUL. Consistently,with REUL expression, the ubiquitination levels of wild type RIG-I, K18R and K190R/193R were markedly increased, whereasthose of K154R, K164R, or K172R, the double mutants K164/154R, K172/154R, K172/164R, or the triple mutant K154R/K164R/K172R were minimal. We also detected the similar resultwithout REUL expression (Fig. 6C). These data suggest that Lys154, 164, and 172 are the primary sites for efficient REUL-mediated ubiquitination of RIG-I.

Discussion

REUL has also been identified as Riplet in a recent publication[20]. Riplet has been reported to promote RIG-I-mediatedinterferon-b promoter activation, and inhibit propagation of anegative-strand RNA virus (vesicular stomatitis virus). Our resultsconfirmed the functions of this protein. But there are severalcritical differences between the findings for REUL and Riplet.Firstly, Riplet was reported to promote lysine 63-linked poly-ubiquitination of the C-terminal region of RIG-I, whichmodification was considered by the authors to differ from the N-terminal ubiquitination by TRIM25 [16]. But in our study, noapparent ubiquitination level was detected in overexpression ofRIG-IDCARD (carrying helicase and RD domains) alone ortogether with REUL. As a control, in the same conditions, full-length RIG-I was extensively ubiquitinated and REUL expressionmarkedly increased the ubiquitination levels of RIG-I. Theseresults suggested that the CARD domain, but not the helicasedomain of RIG-I is ubiquitinated through REUL action.Secondly, Riplet has been reported to bind to the C-terminal

helicase and RD regions of RIG-I, and fails to co-precipitate theCARD domains of RIG-I [16]. In our study, the CARD andhelicase domains of RIG-I interacted with REUL individually, andefficiently as full-length RIG-I. This result suggested that there aremultiple binding sites on RIG-I which mediate the interaction withREUL.

Furthermore, Riplet has been reported to potentiate polyI:C-induced activation (cells stimulated with medium containingpolyI:C) of the IFN-b promoter [16]. In our study, polyI:C didnot activate IFN-b expression in TLR3-deficient 293T cells whenapplied extracellularly (data not shown). And REUL or RE-ULDRING also had no effect on signaling induced by intracellular(transfected) polyI:C in HEK293T cells. Previously, it has beendemonstrated that, in mouse embryonic fibroblasts, MDA5 isinvolved in IFN induction triggered by polyI:C transfected into thecytoplasm [3]. And in reporter assays, knockdown of MDA5significantly inhibits cytoplasmic polyI:C-induced ISRE activation,whereas knockdown of RIG-I has a minimal inhibitory effect [19],suggesting that cytoplasmic polyI:C-induced IRF-3 activation ismostly mediated by MDA5 in HEK293 cells. In our study, REULwas associated with RIG-I but not MDA5 in mammalian cells(Fig. 2). Overexpression of REUL inhibited RIG-I- but notMDA5-mediated activation of ISRE, NF-kB and IFN-b promoter(Fig. 3). Overexpression of REULDRING efficiently inhibited SV,but not polyI:C-induced IFN-b promoter activation in HEK293Tcells (Fig. 4). These results confirm that REUL is specificallyinvolved in the RIG-I-, but not the MDA5-mediated signalingpathway.Another E3 ligase, Trim25, has recently been shown to induce

K63-linked ubiquitination of RIG-I and to contribute to theactivation of downstream signaling pathways [15]. TRIM25 andREUL have homologous domain patterns and regulate RIG-I in asimilar manner. Both of them deliver ubiquitin to the CARDdomain of RIG-I. In addition, REUL interacts with RIG-Ithrough its C-terminal PRY and SPRY domains; this is the sameas TRIM25, which also contains a C-terminal SPRY domain. Onthe other hand, the CARD domain of RIG-I has been reported tobind TRIM25, and the interaction between the helicase domainand TRIM25 has not been identified [15]. In our study, we foundthat in the same way, REUL and TRIM25 interacted with thehelicase domain as well as the CARD domain of RIG-I. Theseresults suggested that there are multiple binding sites on RIG-Iwhich mediate the interaction with REUL and TRIM25, althoughonly the CARD domain is their ubiquitination target.The Lys 172 residue of RIG-I is critical for efficient TRIM25-

mediated ubiquitination and for VISA binding [15], as well as theability of RIG-I to induce antiviral signal transduction. In the samereport, mutations of K99, 169, 181, 190 and 193 weredemonstrated to induce IFN-b and NF-kB promoter efficiently.Our study confirmed these results and notably suggested that Lys154 and Lys 164 are also critical sites for RIG signaling. TheK154R and K164R mutants disrupted the signaling function ofRIG-I to induce activation of IFN-b promoter as strongly as theK172R mutant. Combined double or triple mutants caused near-complete loss of the ability of RIG-I to induce activation of IFN-bpromoter. Our results showed that not only K172, but also K154and K164 are involved in REUL-mediated RIG-I ubiquitinationand signaling.Although TRIM25 and REUL regulate RIG-I in a similar

manner, they function independently of each other. Using Trim25siRNA, we found that in TRIM25 knock-down cells, REUL

HEK293-RIG-I stable cell line and HEK293-MDA5 stable cell line (26105 cells) were transfected with indicated plasmids. 24 h after transfection,immunofluorescent staining was performed with anti-HA antibody (red) and anti-IRF3 antibody (green). The cells expressed REUL or VISA wascounted, and the percentage of cells with nuclear IRF3 was calculated. The experiments were repeated twice, and similar results were obtained. (F)REUL potentiates RIG-I-, but not MDA5-mediated gene expression. HEK293T cells (26105) were transfected with indicated plasmids. 24 h aftertransfection, total RNA was isolated and RT-PCR was performed using indicated primers. (G), (H), (I) RING domain deletion mutant REULDRINGsuppresses RIG-I-, but not MDA5-mediated activation of ISRE (H), NF-kB (I) and IFN-b (G) promoter in a dose-dependent manner. Transfection andluciferase assay were performed as in (A), (B) and (C).doi:10.1371/journal.pone.0005760.g003



Figure 4. REUL potentiates RIG-I antiviral activity. (A), (B), (C) RING domain deletion mutant REULDRING suppresses SV-induced activation ofISRE (A), NF-kB (B) and IFN-b (C) promoter in a dose-dependent manner. HEK293T cells (16105) were transfected with the indicated reporter plasmid(50 ng), pRL-TK Renilla luciferase plasmid (50 ng) and the indicated amounts of an expression plasmid for REULDRING or empty plasmids. 24 h aftertransfection, cells were left uninfected or infected with SV for 8 h before luciferase assays were performed. (D) REULDRING has no effect on polyI:C-induced IFN-b promoter. The transfections were done as in (A), (B) and (C). 16 h after transfection, cells were further transfected with polyI:C (4 mg) oruntransfected for 12 h before luciferase assays were performed. (E) Effects of REUL RNAi plasmids on the expression of transfected REUL. HEK293Tcells (26105) were transfected with expression plasmids for HA-REUL and HA-WDR34 as control (0.5 mg each), and the indicated RNAi plasmids (1 mg).At 48 h after transfection, cell lysates were analyzed by Western blot with anti-HA and anti-GAPDH antibody. (F), (G), (H) REUL RNAi plasmids suppress



SV-induced activation of ISRE (F), NF-kB (G) and IFN-b (H) promoter. HEK293T cells (16105) were transfected with the indicated reporter plasmid(50 ng), pRL-TK Renilla luciferase plasmid (50 ng), and the indicated REUL RNAi (1 mg). At 40 h after transfection, cells were left uninfected or infectedwith SV for 8 h before luciferase assays were performed. (I) REUL RNAi suppresses RIG-I-, but not MDA5-mediated gene expression. HEK293T cells(26105) were transfected with indicated plasmids. At 40 h after transfection, cells were left uninfected or infected with SV for 8 h before total RNAwas isolated and RT-PCR was performed using indicated primers. (J) REUL suppresses NDV-eGFP replication. HEK293T cells (16105) were transfectedwith the indicated plasmids. At 20 h after transfection, cells were infected with NDV-eGFP at MOI 0.001. At 40 h after infection, virus titer andreplication were determined by plaque assay or GFP expression visualized by fluorescence microscopy. Pfu, plaque-forming unit. (K) REUL acceleratesRIG-I-mediated anti-VSV response. HEK293T cells (26105) were transfected with indicated plasmids (0.5 mg each). At 30 h after transfection, cells wereinfected with VSV (MOI = 0.001) and supernatants were harvested at 12, 18, 24 and 30 h post infection. Supernatants were analyzed for VSVproduction using standard plaque assays. Plaques were counted and titers calculated as plaque-forming units (pfu/ml). (L) Knockdown of REULinhibits RIG-I-mediated anti-VSV response. The experiments were carried out as in (K).doi:10.1371/journal.pone.0005760.g004

Figure 5. REUL is an E3 ubiquitin ligase of RIG-I. (A) REUL increases the ubiquitination levels of RIG-I in overexpression system. HEK293T cellswere transfected with indicated REUL, REULDRING, RIG-I or ubiquitin plasmids. Cell lysates were immunoprecipitated (IP) with anti-HA (aHA). Theimmunoprecipitates were analyzed by Western blot (IB) with anti-Flag antibody (aF). Whole-cell lysates were analyzed by Western blots with anti-HAor anti-Flag to determine the expression of REUL, REULDRING and RIG-I. (B) REUL increases the ubiquitination levels of RIG-I in TRIM25 knock downcells. HEK293T cells were transfected with TRIM25 RNAi and indicated RIG-I, REUL and REULDRING plasmids. 48 h after transfection, cells were lysated.Immunoprecipitation and western blot were performed as in (A). (C) REUL delivered ubiquitin to RIG-I in vitro. In vitro expressed REUL and RIG-I wereused to ubiquitination assay, Biotinylated ubiquitin was detected by HRP-streptavidin. (D) REUL effects on ubiquitination of endogenous RIG-I.HEK293T cells transfected with indicated REUL, REULDRING, and ubiquitin plasmids. 24 h after transfection, cells were left uninfected or infected withSV for 24 h before lysate. Immunoprecipitation and Western blot were performed as in (A) except that anti-RIG-I antibody (aRIG-I) were used todetect endogenous RIG-I. (E) REUL does not increase the ubiquitination levels of RIG-I DCARD. HEK293T cells transfected with indicated REUL, RIG-I,RIG-I DCARD and ubiquitin plasmids. Cell lysates were immunoprecipitated with anti-HA (aHA). The immunoprecipitates were analyzed by Westernblot with anti-Flag antibody (aF). Whole-cell lysates were analyzed by Western blot with anti-HA or anti-Flag to determine the expression of REUL,RIG-I and RIG-IDCARD.doi:10.1371/journal.pone.0005760.g005



Figure 6. Lys 154, 164, 172 are the essential sites for RIG-I and REUL signaling. (A) K154R, K164R and K172R are the essential sites for RIG-Iinduced activation of IFN-b promoter. HEK293T cells (16105) were transfected with IFN-b promoter reporter plasmid (50 ng), pRL-TK Renilla luciferaseplasmid (50 ng), and the indicated expression plasmids of RIG-I wild type (WT) and mutations. Reporter assays were performed 24 h aftertransfection. Lower panel: expression of RIG-I wild type (WT) and mutations. (B) K154R, K164R and K172R are the essential sites for REUL effect.Transfection and luciferase assay were performed as in (A), except cotransfection with REUL. (C) K154R, K164R and K172R are the essential sites forREUL-mediated RIG-I ubiquitination. HEK293T cells were transfected with indicated expression plasmids of RIG-I wild type, mutations and ubiquitin,



facilitated the ubiquitination of RIG-I efficiently. And in vitroubiquitin assays showed that REUL directly delivered ubiquitin toRIG-I. These results demonstrated that the function of REUL isindependent of TRIM25. TRIM25 and REUL are not ubiqui-tously distributed over the organs tested. Northern blots showedthat human Riplet is expressed in human skeletal muscle, spleen,kidney, placenta, prostate, stomach, thyroid and tongue, andweakly expressed in heart, thymus, liver, lung and uterus [16].Whereas TRIM25/Efp is predominantly expressed in variousfemale organs such as the uterus, ovary, mammary gland andplacenta [20]. Since TRIM25 and REUL have differentdistribution patterns, we suggest that REUL may be a complementfactor of TRIM25 for RIG-I activation.We noted that expression of the mutant with the RING domain

deleted, REULDRING, markedly suppressed RIG-I-mediatedand SV-induced ISRE, NF-kB and IFN-b promoter, and alsoefficiently inhibited NDV replication and RIG-I ubiquitination.These data indicate that, as a dominant negative mutant,REULDRING competes with endogenous REUL to bind withand ubiquitinate RIG-I, and block RIG-I signaling.As a cytosolic receptor, RIG-I activity is tightly regulated by

several mechanisms. LGP2, a RNA helicase protein lackingCARD domains at its N terminus, acts as a negative regulator bysequestering viral RNA from RIG-I [21,22]. CYLD targets RIG-Ifor deubiquitination that leads to inactivation of the signaling[23,24]. The autophagic protein conjugate Atg5-Atg12 negativelyregulates the type I IFN production pathway by direct associationwith the CARD domain of RIG-I [25]. ISG15 has been reportedto modify RIG-I and act as a negative feedback regulator [26].Two RING finger proteins, RNF125 and TRIM25, have beenshown to play critical roles in regulating RIG-I activity as E3ubiquitin ligase. RNF125-mediated K48-linked polyubiquitinationleads to the degradation of RIG-I [13], and TRIM25-mediatedK63-linked polyubiquitination activates RIG-I [15]. In the presentwork, we identified REUL as a novel RIG-I E3 ubiquitin ligasethat specifically potentiates RIG-I-mediated antiviral activity. Thisfinding extends the regulator family of RIG-I, and the mechanismsunderlying this finding deserve further study, especially usingknockout techniques.

Materials and Methods

Reagents and cell linesMouse monoclonal antibodies against Flag and HA epitopes

(Sigma, USA), poly(I:C) (InvivoGen, USA), IRDye800-conjugatedanti-mouse and anti-rabbit IgG (Rockland Immunochemicals,USA), mouse monoclonal antibody to RIG-I (Alexis Biochemicals,Switzerland), rabbit polyclonal antibodies against IRF3 (SC-9082),SV (Hong-Bing Shu, Wuhan University, Wuhan, China), NDV-eGFP (Cheng Wang, Institute of Biochemistry and Cell Biology,Shanghai, China) and VSV (Hong-Kui Deng, Peking University,Beijing, China) were obtained from the indicated sources. Mouseanti-REUL antibody was prepared with recombinant protein byExperimental Animal Center, Institute of Genetics and Develop-mental Biology, Chinese Academy of Sciences. HEK293, BHK21,HT1080, Hela, HUVEC and U937 cells were grown in DMEMsupplemented with 10% fetal bovine serum. K562 cells weregrown in IMDM supplemented with 10% fetal bovine serum.

Yeast Two-Hybrid ScreensThe human fetal kidney cDNA library (Clontech, USA) was

screened with full-length RIG-I as bait, following protocolsrecommended by the manufacturer.

ConstructsMammalian expression plasmids for Flag- or HA-tagged

TRIM25, TRIM-5a, RIG-I, REUL, and their deletion mutantswere constructed by standard molecular biology techniques.Mammalian expression plasmids for Flag-VISA, Flag-MDA5,Flag-Ub, and ISRE, NF-kB, and IFN-b promoter luciferasereporter plasmids were kindly provided by Dr. Hong-Bing Shu(Wuhan University, Wuhan, China).

Transfection and Luciferase AssaysHEK293T cells were seeded on 24-well dishes and transfected

the next day by standard calcium phosphate precipitation. In thesame experiment, we added empty control plasmid to ensure thateach transfection received the same amount of total DNA. Tonormalize for transfection efficiency, 0.05 mg of pRL-TK (Renillaluciferase) reporter plasmid was added to each transfection.Luciferase assays were performed using a dual-specific luciferaseassay kit (Promega, USA). Firefly luciferase activity was normal-ized based on Renilla luciferase activity. All reporter assays wererepeated at least three times. Data shown are average values 6SDfrom one representative experiment.

Coimmunoprecipitation and Western Blot AnalysisFor transient transfection and immunoprecipitation experi-

ments, HEK293T cells (,26105) were transfected with theindicated plasmids for 20 h. The transfected cells were lysed in0.5 ml of lysis buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 1%Triton X-100, 1 mM EDTA, 10 mg/ml aprotinin, 10 mg/mlleupeptin, 1 mM phenylmethylsulfonyl fluoride). For each immu-noprecipitation, a 0.4 ml aliquot of lysate was incubated with0.5 mg of the indicated antibody and 25 ml of a 1:1 slurry ofprotein-A sepharose (GE Healthcare, USA) for 2 h. The sepharosebeads were washed three times with 1 ml of lysis buffer. Theprecipitates were analyzed by Western blot with the indicatedantibodies and visualized by incubation with IRDye800-conjugat-ed secondary antibodies (diluted 1:10,000) using an Odysseyinfrared imaging system (LICOR Inc., Germany).

Native PAGENative PAGE (8%) was pre-run for 30 min at 40 mA with

native running buffer (25 mM Tris and 192 mM glycine, pH 8.4)and with 0.5% deoxycholate in the cathode chamber. Sampleswere prepared in the native sample buffer (62.5 mM Tris–HCl,pH 6.8, 15% glycerol and 0.5% deoxycholate).

Immunofluorescent stainingCells were fixed with ice-cold methanol for 10 min at 220uC,

rehydrated three times with PBS, and blocked in 5% bovine serumalbumin/PBS for 10 min. The cells were stained with primaryantibody in blocking buffer for 1 h at 37uC, rinsed with PBS, andstained again with FITC-conjugated Affinipure rabbit anti-mouseIgG or Texas Red-conjugated Affinipure goat anti-rabbit IgG

together with or without REUL expression plasmid. Cell lysates were immunoprecipitated with anti-HA (aHA). The immunoprecipitates (IP) wereanalyzed by Western blot (IB) with anti-Flag antibody (aF). Whole-cell lysates (WCL) were analyzed by Western blots with anti-HA or anti-Flag todetermine the expression of REUL, RIG-I wild type and mutations.doi:10.1371/journal.pone.0005760.g006



(1:200 dilutions) for 1 h at 37uC. The cells were then rinsed withPBS containing DAPI and mounted in Prolong Antifade(Molecular Probes). The cells were observed under an OlympusBX51 immunofluorescence microscope using a 6100 planobjective.

In vitro Ubiquitinlation AssayIn vitro protein expression was performed using TNTH quick

coupled transcription/translation systems (Promega, USA), Ubi-quitinlation Assay was performed using ubiquitinylation kit(ENZO life sciences, USA), Following protocols recommendedby the manufacturer.

RNAi ExperimentsDouble-strand oligonucleotides corresponding to the target

sequences were cloned into the pSuper.retro RNAi plasmid(Oligoengine, USA). In this study, the target sequences for humanREUL cDNA were 1#: 59AACATCTTGTAGACATTGTCA39; 2#: 59AATCAGAGATATTCTCCATGA 39; 3#:59AAGTG-AAGTGGACACTAGGAATTGCA 39; positive control VISARNAi target sequence 59GTATATCTGCCGCAATTTC 39.

RT-PCRTotal RNA was isolated from HEK293T cells by using TRIzol

reagent (Tianwei Co., Beijing, China) and subjected to RT-PCRanalysis to measure expression of IFN-b, ISG15, RANTES, TNFaand b-actin. The gene-specific primer sequences were: IFN-b, sense:59-CCAACAAGTGTCTCCTCCAA-39, antisense: 59-ATAGTCT-CATTCCAGCCAGT-39; ISG15, sense: 59-GCTGGGACCT-GACGGTGAAGA-39, anti-sense: 59-GCTCAGAGGTTCGTCG-CATTTGT-39; RANTES, sense: 59-CCTCGCTGTCATCC-TCATTG-39, anti-sense: 59-TACTCCCGAACCCATTTCTT-39;

TNFa, sense: 59-CCTGGTATGAGCCCATCTATC-39, antisense:59-CGAAGTGGTGGTCTTGTTGC-39; b-Actin, sense: 59-ACGTGGACATCCGCAAAGAC-39, antisense:59-CAAGAAAG-GGTGTAACGCAACTA-39.

VSV plaque assayHEK293T cells (,16105) were transfected with the indicated

plasmids for 24 h prior to VSV infection. At 1 h post-infection,cells were washed with PBS and then fresh medium was added.The supernatant was harvested at the indicated times and used toinfect confluent BHK21 cells cultured on 24-well plates. At 1 hpost-infection, supernatant was removed and culture mediumcontaining 2% methylcellulose was overlaid. At 60 h post-infection, the overlaid medium was removed; cells were fixedwith 0.5% glutaraldehyde for 30 min and stained with 1% crystalviolet dissolved in 70% ethanol. Plaques were counted, averagedand multiplied by the dilution factor to determine viral titer asPfu/ml. The experiments were repeated three times and eachexperiment was performed in duplicate. Data shown are averagevalues 6S.D. from one representative experiment.

Acknowledgments

We thank Dr. Hong-Bing Shu for stimulating discussion and members ofour laboratory for technical help.

Author Contributions

Conceived and designed the experiments: DG DC. Performed theexperiments: DG YY RW XZ FD ML DC. Analyzed the data: DG DC.Contributed reagents/materials/analysis tools: ZZ ZJ DC. Wrote thepaper: DC.

References

1. Takeuchi O, Akira S (2007) Recognition of viruses by innate immunity.Immunol Rev 220: 214–224.

2. Wathelet MG, Lin CH, Parekh BS, Ronco LV, Howley PM, et al. (1998) Virusinfection induces the assembly of coordinately activated transcription factors onthe IFN-beta enhancer in vivo. Mol Cell 1: 507–518.

3. Kato H, Takeuchi O, Sato S, Yoneyama M, Yamamoto M, et al. (2006)Differential roles of MDA5 and RIG-I helicases in the recognition of RNAviruses. Nature 441: 101–105.

4. Kato H, Sato S, Yoneyama M, Yamamoto M, Uematsu S, et al. (2005) Celltype-specific involvement of RIG-I in antiviral response. Immunity 23: 19–28.

5. Hornung V, Ellegast J, Kim S, Brzozka K, Jung A, et al. (2006) 59-TriphosphateRNA is the ligand for RIG-I. Science 314: 994–997.

6. Pichlmair A, Schulz O, Tan CP, Naslund TI, Liljestrom P, et al. (2006) RIG-I-mediated antiviral responses to single-stranded RNA bearing 59-phosphates.Science 314: 997–1001.

7. Fitzgerald KA, McWhirter SM, Faia KL, Rowe DC, Latz E, et al. (2003)IKKepsilon and TBK1 are essential components of the IRF3 signaling pathway.Nat Immunol 4: 491–496.

8. Xu LG, Wang YY, Han KJ, Li LY, Zhai Z, et al. (2005) VISA is an adapterprotein required for virus-triggered IFN-beta signaling. Mol Cell 19: 727–740.

9. Meylan E, Curran J, Hofmann K, Moradpour D, Binder M, et al. (2005) Cardifis an adaptor protein in the RIG-I antiviral pathway and is targeted by hepatitisC virus. Nature 437: 1167–1172.

10. Kawai T, Takahashi K, Sato S, Coban C, Kumar H, et al. (2005) IPS-1, anadaptor triggering RIG-I- and Mda5-mediated type I interferon induction. NatImmunol 6: 981–988.

11. Sharma S, tenOever BR, Grandvaux N, Zhou GP, Lin R, et al. (2003)Triggering the interferon antiviral response through an IKK-related pathway.Science 300: 1148–1151.

12. Seth RB, Sun L, Ea CK, Chen ZJ (2005) Identification and characterization ofMAVS, a mitochondrial antiviral signaling protein that activates NF-kappaBand IRF 3. Cell 122: 669–682.

13. Arimoto K, Takahashi H, Hishiki T, Konishi H, Fujita T, et al. (2007) Negativeregulation of the RIG-I signaling by the ubiquitin ligase RNF125. Proc NatlAcad Sci U S A 104: 7500–7505.

14. Chen ZJ (2005) Ubiquitin signalling in the NF-kappaB pathway. Nat Cell Biol 7:758–765.

15. Gack MU, Shin YC, Joo CH, Urano T, Liang C, et al. (2007) TRIM25 RING-finger E3 ubiquitin ligase is essential for RIG-I-mediated antiviral activity.Nature 446: 916–920.

16. Oshiumi H, Matsumoto M, Hatakeyama S, Seya T (2009) Riplet/RNF135, aRING finger protein, ubiquitinates RIG-I to promote interferon-beta inductionduring the early phase of viral infection. J Biol Chem 284: 807–817.

17. Douglas J, Cilliers D, Coleman K, Tatton-Brown K, Barker K, et al. (2007)Mutations in RNF135, a gene within the NF1 microdeletion region, causephenotypic abnormalities including overgrowth. Nat Genet 39: 963–965.

18. Jenne DE, Tinschert S, Reimann H, Lasinger W, Thiel G, et al. (2001)Molecular characterization and gene content of breakpoint boundaries inpatients with neurofibromatosis type 1 with 17q11.2 microdeletions. Am J HumGenet 69: 516–527.

19. Jenne DE, Tinschert S, Dorschner MO, Hameister H, Stephens K, et al. (2003)Complete physical map and gene content of the human NF1 tumor suppressorregion in human and mouse. Genes Chromosomes Cancer 37: 111–120.

20. Orimo A, Inoue S, Ikeda K, Noji S, Muramatsu M (1995) Molecular cloning,structure, and expression of mouse estrogen-responsive finger protein Efp. Co-localization with estrogen receptor mRNA in target organs. J Biol Chem 270:24406–24413.

21. Rothenfusser S, Goutagny N, DiPerna G, Gong M, Monks BG, et al. (2005) TheRNA helicase Lgp2 inhibits TLR-independent sensing of viral replication byretinoic acid-inducible gene-I. J Immunol 175: 5260–5268.

22. Yoneyama M, Kikuchi M, Matsumoto K, Imaizumi T, Miyagishi M, et al.(2005) Shared and unique functions of the DExD/H-box helicases RIG-I,MDA5, and LGP2 in antiviral innate immunity. J Immunol 175: 2851–2858.

23. Friedman CS, O’Donnell MA, Legarda-Addison D, Ng A, Cardenas WB, et al.(2008) The tumour suppressor CYLD is a negative regulator of RIG-I-mediatedantiviral response. EMBO Rep 9: 930–936.

24. Zhang M, Wu X, Lee AJ, Jin W, Chang M, et al. (2008) Regulation of IkappaBkinase-related kinases and antiviral responses by tumor suppressor CYLD. J BiolChem 283: 18621–18626.

25. Jounai N, Takeshita F, Kobiyama K, Sawano A, Miyawaki A, et al. (2007) TheAtg5 Atg12 conjugate associates with innate antiviral immune responses. ProcNatl Acad Sci U S A 104: 14050–14055.

26. Kim MJ, Hwang SY, Imaizumi T, Yoo JY (2008) Negative feedback regulationof RIG-I-mediated antiviral signaling by interferon-induced ISG15 conjugation.J Virol 82: 1474–1483.



Riplet/RNF135, a RING Finger Protein, Ubiquitinates RIG-I toPromote Interferon-! Induction during the Early Phase ofViral Infection*□S

Received for publication, June 3, 2008, and in revised form, November 10, 2008 Published, JBC Papers in Press, November 18, 2008, DOI 10.1074/jbc.M804259200

Hiroyuki Oshiumi‡, Misako Matsumoto‡, Shigetsugu Hatakeyama§, and Tsukasa Seya‡1

From the ‡Department of Microbiology and Immunology and the §Department of Biochemistry, Hokkaido University GraduateSchool of Medicine, Kita-15, Nishi-7, Kita-ku Sapporo 060-8638, Japan

RIG-I (retinoic acid-inducible gene-I), a cytoplasmic RNAhelicase, interacts with IPS-1/MAVS/Cardif/VISA, a protein onthe outer membrane of mitochondria, to signal the presence ofvirus-derived RNA and induce type I interferon production.Activation of RIG-I requires the ubiquitin ligase, TRIM25,whichmediates lysine 63-linkedpolyubiquitinationof theRIG-IN-terminal CARD-like region. However, how this modificationproceeds for activation of IPS-1 by RIG-I remains unclear. Herewe identify an alternative factor, Riplet/RNF135, that promotesRIG-I activation independent of TRIM25. The Riplet/RNF135protein consists of anN-terminal RING finger domain,C-termi-nal SPRY and PRY motifs, and shows sequence similarity toTRIM25. Immunoprecipitation analyses demonstrated that theC-terminal helicase and repressor domains of RIG-I interactwith the Riplet/RNF135 C-terminal region, whereas the CARD-like region of RIG-I is dispensable for this interaction. Riplet/RNF135 promotes lysine 63-linked polyubiquitination of theC-terminal region of RIG-I, modification of which differs fromthe N-terminal ubiquitination by TRIM25. Overexpression andknockdown analyses revealed that Riplet/RNF135 promotes RIG-I-mediated interferon-! promoter activation and inhibits propa-gation of the negative-strandRNAvirus, vesicular stomatitis virus.Our data suggest that Riplet/RNF135 is a novel factor of the RIG-Ipathway that is involved in the evoking of human innate immunityagainst RNA virus infection, and activates RIG-I through ubiquiti-nation of its C-terminal region. We infer that a variety of RIG-I-ubiquitinating molecular complexes sustain RIG-I activation tomodulate RNA virus replication in the cytoplasm.

Cytoplasmic viral RNA sensors induce production of type Iinterferon (IFN)2 (1, 2). Representative cytoplasmic sensors,

RIG-I-like receptors (RLRs) of RIG-I,MDA5, and LGP2, belongto the DEA(D/H) box RNA helicase family (3–6). RIG-I recog-nizes the 5! end triphosphate of the virus RNA genome or dou-ble-stranded RNA (6–8) to sense infection by various RNAviruses (3, 5). The RIG-I protein consists of two N-terminalCARD-like domains, an RNA helicase region and a repressordomain (RD) (9). After recognition of positive or negative sin-gle-stranded viral RNA, RIG-I interacts with its adaptor mole-cule IPS-1/MAVS/Cardif/VISA leading to type I IFN produc-tion, thereby protecting host cells from amplified viralreplication (10–13). However, only a few copies of viral RNAsusually penetrate the cell membrane to enter the cell at an earlyinfection, and these RLRs are barely present in intact as well asearly virus-infected cells (6). The early viral RNA recognitionfacility should be different from that of the late phase whenRIG-I protein is abundant in the cytoplasm and easily re-orga-nizes the virus RNAs. What molecular mechanism is responsi-ble for initial sensing of viral RNA thus remains unknown.Other RLRs, MDA5 and LGP2, are structurally similar to

RIG-I in their having the helicase domain (5, 14). However,MDA5 lacks the RD domain although it possesses CARD-likeregion at the N terminus like RIG-I. LGP2 does not have aCARD-like region but possesses RD at its C terminus (9). RIG-Iand MDA5 recognize different kinds of RNA viruses and insome cases play a redundant role in sensing virus infection,such as influenza B (15). In contrast, LGP2 rather negativelyregulates virus replication. LGP2 expression suppressed RIG-IorMDA5 signaling (14, 16), and lgp2 gene disruption conferredhigh susceptibility to virus infection on mice (4).Recently, the majority of proteins involved in the type I IFN-

inducing system were found ubiquitinated. For example, thetumor necrosis factor receptor-associated family members,TRAF3 and TRAF6, are ubiquitin ligases to induce ubiquitina-tion of proteins and implicated in activation of IFN regulatoryfactor (IRF) 3 or nuclear factor (NF) !B (13, 17–19). In contrast,a deubiquitinating enzyme, DUBA or A20, suppresses thesesignals (19, 20). In addition to ubiquitin, ubiquitin-like protein,ISG15, is also conjugated to proteins involved in the IFN-induc-ing pathway (21, 22). Recent studies have revealed that viralRNA sensors are also ubiquitinated. TRIM25 (ZNF147 or EFP),a member of the ubiquitin-protein isopeptide ligase family,which possesses a RING finger domain, ubiquitinates the

* This work was supported in part by grants-in-aid from the Ministry of Education,Science and Culture of Japan, Ministry of Health, Labour, and Welfare, TheMitsubishi Foundation, and The Mochida Memorial Foundation. The costs ofpublication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked “advertisement” inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

□S The on-line version of this article (available at http://www.jbc.org) containssupplemental Figs. S1–S6.

The nucleotide sequence(s) reported in this paper has been submitted to the Gen-BankTM/EBI Data Bank with accession number(s) AB470605.

1 To whom correspondence should be addressed: Dept. of Microbiology andImmunology, Graduate School of Medicine, Hokkaido University, Kita-ku,Sapporo 060-8638, Japan. Tel.: 81-11-706-5073; Fax: 81-11-706-7866;E-mail: [email protected].

2 The abbreviations used are: IFN, interferon; RT, reverse transcription; RLR,RIG-I-like receptor; HA, hemagglutinin; siRNA, small interference; m.o.i.,

multiplicity of infection; VSV, vesicular stomatitis virus; IRF, IFN regulatoryfactor; Ub, ubiquitin; ORF, open reading frame; RD, repressor domain.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 284, NO. 2, pp. 807–817, January 9, 2009© 2009 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

JANUARY 9, 2009 • VOLUME 284 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 807

CARD-like domains of RIG-I thereby facilitating the RIG-I-mediated activation of type I IFN signaling (23, 24), althoughShimotohno and co-workers (25) previously reported thatTRIM25 (EFP) does not polyubiquitinate the RIG-I CARD-likeregion as far under their conditions. Expression of TRIM25increases RIG-I CARD-like region-mediated signaling; how-ever, it remains to be determined whether the activation offull-length RIG-I requires other ubiquitin ligase (23). Anotherubiquitin ligase RNF125 mediates lysine 48-linked polyubiq-uitination of RIG-I, which leads to degradation of RIG-Ithrough the proteasome (25).Here we examined what molecular complex participates in

an early RIG-I-mediated RNA recognition and IFN signaling byyeast two-hybrid screening. Here we detected two novel RINGfinger proteins that bound to RIG-I, and we found that one,RNF135, facilitated RIG-I-mediated type I IFN induction viaubiquitinating RIG-I. RNF135 plays a crucial role in the RIG-Iresponse to minimal copies of viral RNA, and by binding to theC-terminal helicase and RD regions of RIG-I, RNF135 facili-tates RIG-I C-terminal ubiquitination to up-regulate RIG-I-mediated IFN signaling and suppress viral replication. Hence,we renamed it as RNF135 Riplet (RING finger protein leadingto RIG-I activation). To our knowledge, this is the first studydemonstrating that C-terminal ubiquitination of RIG-I isimportant for full IFN induction by RIG-I.

EXPERIMENTAL PROCEDURES

Cell Cultures—HEK293 and Vero cells were cultured in Dul-becco’s modified Eagle’s medium with 10% fetal calf serum(Invitrogen), and HeLa cells were in minimum Eagle’s mediumwith 2 mM L-glutamine and 10% fetal calf serum (JRH Bio-sciences). HEK293FT cells were maintained in Dulbecco’smodified Eagle’s high glucose medium containing 10% heat-inactivated fetal calf serum (Invitrogen).Plasmids—cDNA fragment encoding a C-terminal region of

Riplet was isolated by yeast two-hybrid screening using humanlung cDNA library. The 5! region encoding the remainingN-terminal region was amplified by PCR using primers Rip-let-F1 and Riplet-R1, and human lung cDNA library was usedfor its template. Two cDNA fragments, which cover the entireORF of Riplet, were joined by PCR using primers Riplet-F1, R1,F2, and R2 and then inserted into pCR-blunt vector (Invitro-gen). The primers sequences are as follows: F1, GCCTCGAG-GCCACCATGGCGGGCCTGGGCCTGGG; R1, CGGCCAG-GTCCTGCAGTAGC; F2, GCACCTGCGGAAGAACACGC;and R2, GGGGATCCCACCTTTACTTGCTTTATTATC-AGG. The obtained cDNA was cloned into XhoI-NotI restric-tion sites of pEF-BOS expression vector, and the HA tag wasfused at the C-terminal end of Riplet. Riplet-DN (dominantnegative) expression vector was constructed by amplifying therelevant Riplet cDNA fragment using the primers Riplet-X-F-Cand Riplet-R2 and subcloned into pEF-BOS. The primersequence of Riplet-X-F-C was as follows: GCTCGAGGCCAC-CATGCCGCACCTGCGGAAGAACACGC. Riplet-L248fs ex-pression vector was made by deleting 1 base at position 742 bystandard PCR-mediated site-directed mutagenesis methodswith primers Riplet-L248fs-F and Riplet-L248fs-R as follows:Riplet-L248fs-F, CCAGAGCCACCCTGCATCAGGAGAGC-

TTCTCGG, and Riplet-L248fs-R, CCGAGAAGCTCTCCTG-ATGCAGGGTGGCTCTGG. All cloned RIPLET cDNA frag-ments were sequenced, and it was confirmed that there were nomutations. Full-length RIG-I expressing vector, Gal4-IRF-3,Gal4-DBD, and p55 UASG-Luc reporter plasmids were giftsfrom Dr. T. Fujita (Kyoto University, Kyoto, Japan). p125 lucreporter plasmidwas a gift fromDr. T. Taniguchi (University ofTokyo, Tokyo, Japan). RIG-I RD expressing vector was madewith primers RIG-I RD-F and RIG-I RD-R; the RIG-I dRDcDNA fragment, which encodes ORF of RIG-I from the 1- to754-amino acid region, was made by using primers RIG-I-(1–754)F and RIG-I-(1–754)R. The obtained cDNA fragmentswere sequenced, and it was confirmed that there were nomuta-tions caused by PCR. The primers sequences are as follows:RIG-I RD-F, GATGATAAAGGTACCACCGGTAGCAAGTGCTTCCTTCTG; RIG-I RD-R, AAGGAAGCACTTGCTACC GGT GGT ACC TTT ATC ATC ATC ATC; RIG-I-(1–754)F, GC AGA GGA AGA GCA AGA TGA TAT CAG GTCCTCAAT CTT C; and RIG-I-(1–754)R, ATTGAGGACCTGATA TCA TCT TGC TCT TCC TCT GCC TC.Northern Blotting—Human RIPLET 1092-bp cDNA frag-

ment (208–1299) was used for the probe for Northern blotting.The Northern blot membranes, human 12-lane MTN blot andMTNblot III, were purchased fromClontech. The homology ofhuman RIPLET and TRIM25 in the probe region was 46%. Weused a stringent condition for Northern blotting to exclude thecross-hybridization between the RIPLET and TRIM25 genes.Briefly, the probe was labeled with ["-32P]dCTP usingRediprime II Random Prime labeling system (GE Healthcare).The labeled probe was hybridized to the membrane withExpressHyb hybridization solution (Clontech) at 68 °C for 1 h.The membrane was washed with washing solution I (2" SSC,0.05% SDS) for 40 min, and then washed with washing solutionII (0.1" SSC, 0.1% SDS) for 40 min. Riplet mRNA bands weredetected with x-ray film.Reporter Gene Analysis—HEK293 cells were transiently

transfected in 24-well plates using FuGENE HD (RocheApplied Science) with expression vectors, reporter plasmids,and internal control plasmid coding Renilla luciferase. Thetotal amounts of plasmids were normalized with empty vector.For poly(I-C) stimulation, 24 h after transfection, cells werestimulated with medium containing poly(I-C) (50 #g/ml) andDEAE-dextran (0.5 mg/ml) for 1 h, and then the medium wasexchanged with normal medium and incubated for an addi-tional 3 h. Cells were lysed with lysis buffer (Promega) andluciferase, and Renilla luciferase activities were measured bythe dual luciferase assay kit (Promega). Relative luciferase activ-ities were calculated by normalizing luciferase activity byRenilla luciferase activity, and dividing the normalized value bycontrol in which only empty vector, reporter, and internal con-trol plasmid were transfected. Values are expressed as meanrelative stimulations # S.D. for a representative experiment,and each was performed three times in duplicate (unless other-wise indicated in the legends).RNA Interference—Reporter and siRNA (20 nM final concen-

tration) for Riplet or control were transfected into HEK293cells with Lipofectamine 2000 (Invitrogen) by the standardmethod described in the manufacturer’s protocol. Empty vec-

A RIG-I Complement Factor, Riplet

808 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 284 • NUMBER 2 • JANUARY 9, 2009

tor was added to normalize the final plasmid amount. 48 h aftertransfection, cells were stimulated with poly(I-C) for 4 h. ForVSV infection, 24 h after transfection, cells were infected withVSV at m.o.i. $ 1, and cell lysate was prepared after 12 h forreporter gene assays. The degree of gene silencing was con-firmed by RT-PCR using RNA extracted from cells 24 h aftertransfection. PCR primers used for the RT-PCR were Riplet-F3(ACTGGGAAGTGGACACTAGG) and Riplet-R3 (ACTCAT-ACAGAAGCTTCTCC). siRNAs were purchased from Funa-koshi Co., Ltd. (Tokyo Japan), and the siRNA sequences of Rip-let siRNA were GACUAUGGACUCUUGUUGUGU (sense)and ACAACAAGAGUCCAUAGUCCU (antisense). ControlsiRNA sequences were CUGUUGGUUUAGUAAGCCUGU(sense) and AGGCUUACUAAACCAACAGUC (antisense).Another siRNA, Riplet si-1, and control negative siRNA

(silencer negative control 1 siRNA,AM4611) were purchased fromApplied Biosystems. siRNA sequen-ces were Riplet si-1 GGGAAAGCU-UGCCUUCUAUdTdT (sense) andAUAGAAGGCAAGCUUUCCCd-TdC (antisense).Virus Preparation and Infection—

VSV Indiana strain and polioviruswere amplified using Vero cells.HEK293 cells were transfected in24-well plates with plasmid encod-ing RIG-I, Riplet, or no insert. 24 hafter transfection, cells were in-fected with viruses for 24 h, and thetiters of virus in culture supernatantwere measured by plaque assayusing Vero cells. For RNA interfer-ence assay, cells were transfectedwith siRNA with Lipofectamine2000. 24 h after transfection, cellswere infected with viruses at m.o.i.$0.001 for 18 h, and the titer in cul-ture supernatant were determinedby plaque assay.Immunoprecipitation—HEK293FT

cells were transfected in 6-well plateswith plasmids encoding FLAG-tagged RIG-I and/or HA-tagged Rip-let. The plasmid amounts were nor-malized by the addition of emptyplasmid. 24 h after transfection, cellswere lysed with lysis buffer (20 mMTris-HCl (pH 7.5), 125 mM NaCl, 1mM EDTA, 10% glycerol, 1%NonidetP-40, 30 mM NaF, 5 mM Na3VO4, 20mM iodoacetamide, and 2 mM phen-ylmethylsulfonyl fluoride), and thenproteins were immunoprecipitatedwith rabbit anti-HA polyclonal(Sigma) or anti-FLAG M2 mono-clonal antibody (Sigma). The precipi-tated samples were analyzed by SDS-

PAGEand stainedwith anti-HA (HA1.1) (Covance) or anti-FLAGM2 monoclonal antibody. For ubiquitination assay of RIG-I, theplasmid encoding two multiple HA-tagged ubiquitins was used.HEK293FT cells were transfected with plasmids encoding FLAG-tagged RIG-I, Riplet, or 2"HA-tagged ubiquitin. 24 h after trans-fection, cells were lysed, and then RIG-I was immunoprecipitatedas described above.The sampleswere analyzedbySDS-PAGEandstained with anti-HA polyclonal antibody (for detection of ubiq-uitination) or anti-FLAG monoclonal antibody (for detection ofRIG-I). Reproducibility was confirmed with additional experi-ments (see supplemental figures).Construction of RIG-I 3KA and 5KA Mutant Genes—The

C-terminal three or five lysine residues were mutated into ala-nines (designated as 3KA and 5KA). RIG-I 3KA has K888A,K907A, and K909A, whereas RIG-I 5KA has K849A, K851A,

FIGURE 1. Isolation of Riplet by yeast two-hybrid screening. A, yeast cells carrying both RIG-I and Riplet cangrow in selective media (SD-WLH, SD-WLHA), whereas yeast cells carrying RIG-I alone only grow in nonselectivemedia (SD-WL), indicating the physical interaction of RIG-I with Riplet. B, human Riplet protein sequence is60.8% identical to human TRIM25. The RING finger domains and SPRY motifs show higher sequence similaritiesbetween the two proteins. aa, amino acids. C, phylogenetic tree constructed by the Neighbor-Joining methodshows that Riplet is similar to TRIM25. h, m, r, or zf represent human, mouse, rat, or zebrafish, respectively. Thenumbers on the node are bootstrap probabilities (n $ 1000). D, HeLa cell, human primary-cultured fibroblastcell, MRC5, or bone marrow-derived mouse dendritic cell (BM-DC) were stimulated with poly(I-C) (50 #g/ml) forindicated hours. Total RNA was extracted with TRIzol reagent, and then RT-PCR was carried out using primersshown under “Experimental Procedures.” GAPDH, glyceraldehyde-3-phosphate dehydrogenase. E, Northernblot membranes containing 1 #g of poly(A)% RNA per lane from human tissues were blotted with human Ripletprobe.



http://www.jbc.org/cgi/content/full/M804259200/DC1

K888A, K907A, and K909A. Themutant rig-I genes were madeby PCR-mediated site-directed mutagenesis. The primers usedfor the PCR were as follows: K907–909A-forward, GTT CAGACA CTG TAC TCG GCG TGG GCG GAC TTT CAT TTTGAG AAG, and K907–909A-reverse, CTT CTC AAA ATGAAA GTC CGC CCA CGC CGA GTA CAG TGT CTG AAC;K888A-forward, GAC ATT TGA GAT TCC AGT TAT AGCAAT TGA AAG TTT TGT GGT GGA GG, and K888A-re-verse, CCT CCA CCA CAA AACTTT CAA TTG CTA TAA CTGGAA TCT CAA ATG TC; K849–851A-forward, GAG TAG ACCACA TCC CGC CCA GCG CAGTTTTCAAGTTTTG, and K849–851A-reverse, CAA AAC TTGAAA ACT GCG CTG GCG CGGGAT GTGGTC TAC TC. PCR wascarried with Pyrobest Taq polymer-ase, and the obtained clones weresequenced to exclude the clonesharboring PCR error. To constructthe plasmid-expressing mutantRIG-I protein, the wild-type RIG-Igene on pEF-BOS vector wasreplaced with the mutant rig-I gene.RealTimePCR—QuantitativePCR

analyses were carried out usingiCycler iQ real time detection systemwith Platinum SYBR Green qPCRSuperMix-UDG reagent (Invitro-gen). Primer sequences for qPCRwere as follows: hGAPDH-qF, GAGTCAACGGATTTGGTCGT, andhGAPDH-qR, TTG ATT TTGGAG GGA TCT CG; hIFN-$-qF,TGG GAG GAT TCT GCA TTACC, and hIFN-$-qR, CAG CATCTG CTG GTT GAA GA; hMx1-qF, ACC ACA GAG GCT CTCAGC AT, and hMx1-qR, CTC AGCTGG TCC TGG ATC TC; andhIFIT-1-qF, GCA GCC AAG TTTTAC CGA AG, and hIFIT-1-qR,CAC CTC AAA TGT GGG CTTTT. Values were expressed as meanrelative stimulations, and for arepresentative experiment from aminimum of three separate exper-iments, each was performed intriplicate.

RESULTS

RIG-I-binding Proteins—To iso-late the proteins that bind to RIG-I,we performed yeast two-hybridscreening using a human lungcDNA library. Using the RIG-I cen-tral region (213–601 amino acids),

we isolated a clone that encoded a partial ORF of a geneexpressed in a dendritic cell line, DC12, whereas theC-terminalregion of RIG-I (557–925 amino acids) resulted in the isolationof two cDNA clones, which encoded partial C-terminal regionsof ZNF598 and RNF135 (Fig. 1A and data not shown). Prelim-inary expression studies showed that the RNF135 segmentaffected the RIG-I IFN-$ inducing activity, whereas the othertwo proteins had no effect (data not shown).We confirmed the



interaction of RIG-I with ZNF598 or RNF135 in HEK293FTcells by immunoprecipitation (data not shown). RNF135 waspreviously annotated by the genome project and was recentlyfound to be a cause of a genetic disease, neurofibromatosis,although its protein function was unknown. We renamed theprotein Riplet (RING finger protein leading to RIG-I activation)based on the following functional analyses. Riplet was mostsimilar to TRIM25 (60.8% sequence homology), in particularbetween their RING finger domains PRY or SPRY (Fig. 1B).Phylogenetic analysis also supported the notion that Riplet wassimilar to TRIM25 (Fig. 1C). Thus, we hypothesized that, likeTRIM25, Riplet is a ubiquitin ligase.Expression of Riplet—RIG-I mRNA is induced by type I IFN

or poly(I-C) stimulation in mammalian cells. Unlike RIG-I,however, Riplet mRNA was basally expressed in HeLa and pri-mary-culturedMRC-5 cells irrespective of stimulation (Fig. 1Dand data not shown). On the other hand, when we treated bonemarrow-derived dendritic cells with poly(I-C), the basal level ofRiplet mRNA was increased by the stimulation (Fig. 1D), sug-gesting that the regulatory mechanism of Riplet expressionsomewhat differs among cell types, and that Riplet is expressedbefore virus infection in some cell types. Next we performedNorthern blotting of human tissue RNA. Riplet mRNA wasdetected as a single band of 2.4 kbp,which is slightly longer thanthe RNF135 cDNA sequence deposited in GenBankTM (acces-sion number AB470605). Human RIPLET is expressed inhuman skeletal muscle, spleen, kidney, placenta, prostate,stomach, thyroid, and tongue and also weakly expressed inheart thymus, liver, and lung (Fig. 1E).Riplet Enhances RIG-I-mediated IFN-$ Induction—At first

we characterized the role of Riplet in RIG-I-mediated IFNinducing signaling by reporter gene analyses. When RIG-I wasexpressed in HEK293 cells, reporter auto-activation wasobserved even in the absence of exogenous stimulation (Fig.2A) as reported previously (25, 26). Stimulation with poly(I-C)further enhanced the promoter. Co-expression of Riplet withRIG-I potentiated activation of the IFN-$ promoter, whereasexpression of Riplet alone resulted in only marginal activation(Fig. 2A). Detection of endogenous IFN-$ mRNA confirmedthat Riplet enhanced RIG-I-mediated activation of IFN-$ tran-scription (supplemental Fig. S1). The enhancing role of Ripletin IFN-$ promoter activation was also supported by activationof IRF-3 and NF-!B by Riplet (Fig. 2, B and C). In contrast,expression of a Riplet partial fragment (Riplet-DN) (70–432

amino acids) that lacked the N-terminal RING finger domainreduced promoter activation (Fig. 2E). The Riplet-L249fsmutant protein, which was isolated from neurofibromatosispatients (27), did not increase the RIG-I-mediated promoteractivation (Fig. 2D). These data indicate that Riplet augmentsRIG-I-mediated IFN-$ promoter activation, and that both theRING finger domain and the C-terminal region encoding theSPRY and PRY motifs are important for its function. Riplet(residues 70–432) acted as a dominant-negative form (hereaf-ter called Riplet-DN) (Fig. 2,E and F, left panel). This functionalfeature of Riplet-DN was confirmed in Fig. 2, B and C, and waslater confirmed throughRIG-I co-precipitation andubiquitina-tion analyses (see Fig. 5C and supplemental Fig. S4C). Expres-sion of Riplet-DNdid not reduceTLR3 orMDA5 signaling (Fig.2E), suggesting that Riplet-DN is specific for RIG-I signaling.Interestingly, the Riplet-DN only partially suppressed the func-tion of the C-terminal deleted RIG-I (dRD), which is a consti-tutively active form (Fig. 2F, right panel), and RIG-I CARD-likeregion (dRIG-I)-mediated signaling in high or low dose trans-fection of dRIG-Iwas barely inhibited by overexpression of Rip-let-DN (Fig. 2F, center panel). These data suggest that Ripletrequires the RIG-I C-terminal domain (RD) and partial helicaseregion to activate RIG-I signaling.Endogenous Riplet Promotes the RIG-I Signaling—We per-

formed Riplet knockdown by siRNA Riplet using Lipo-fectamine 2000 reagents, instead of FuGENE HD, to reveal thefunction of endogenous Riplet. Two siRNAs (Riplet siRNA andRiplet si-1) that target different sites of the Riplet mRNA andtwo control siRNAs were used for knockdown analyses. Thetwo siRNA or control siRNA were co-transfected with HA-tagged Riplet expression vector into HEK293 cells, and after48 h, cell lysate was prepared and analyzed byWestern blottingwith anti-HA antibody detecting Riplet. The two siRNAs tar-geting Riplet abolished exogenously expressed Ripelt-HA, butcontrol siRNA did not (supplemental Fig. S3). Likewise, bothRiplet siRNA and Riplet si-1 specifically down-regulate thelevel of endogenous Riplet mRNA (Fig. 3, A and B).Using the siRNA, we examined whether Riplet knockdown

reduces RIG-I signaling. As expected, RIG-I-mediated IFN-$promoter activation was reduced by Riplet siRNA or Riplet si-1compared with control siRNA (Fig. 3, A and B), indicating thatRiplet is required for full activation of the RIG-I signaling.Vesicular stomatitis virus (VSV) is a negative-stranded RNAvirus that induces IFN-$production viaRIG-I (3). Although the

FIGURE 2. Riplet enhances IFN-! signaling mediated by RIG-I. A, Riplet enhances the promoter activation by RIG-I. HEK293 cells were transfected withplasmids encoding empty vector, RIG-I (0.1 #g) and Riplet (0.025, 0.05, or 0.1 #g) together with p125-luc (IFN-$ promoter) reporter plasmid in 24-well plates.24 h after transfection, the cells were treated with mock or poly(I-C) (50 #g/ml) for 4 h as described under “Experimental Procedures,” and then luciferaseactivities of cell lysates were measured. Closed or open boxes represent poly(I-C) or mock stimulation, respectively. B, to examine the activation of IRF-3, RIG-I (0.1#g), Riplet (0.1 #g), and/or Riplet-DN (0.1 #g),expressing vectors were transfected into HEK293 cells with reporter plasmids, GAL4 fused IRF-3 (0.05 #g), and thep55 UASG-luc reporter gene (0.05 #g), in which luciferase reporter gene is fused downstream of GAL4 protein-binding site, and therefore activated IRF-3promotes the transcription of luciferase reporter gene. The cells were stimulated with poly(I-C) as described above (34). The total amount of transfected DNA(0.5 #g/well) was kept constant by adding empty vector (pEF-BOS). C, HEK293 cells were transfected with RIG-I (0.1 #g), Riplet (0.1 #g), and/or Riplet-DN (0.1#g) expressing vectors together with the NF-!B reporter plasmid (0.1 #g), and 24 h later, the luciferase activities of cell lysates were measured. D, Riplet-L248fs,which lacks the C-terminal region, did not enhance the activation at all. HEK293 cells were transfected with the plasmids expressing wild-type Riplet (0.1 #g)or Riplet-L248fs (0.1 #g) together with RIG-I expressing vector (0.1 #g) and p125-luc reporter (0.1 #g). 24 h after transfection, cell were stimulated with poly(I-C),and the luciferase activities of cell lysates were determined as described above. E, RIG-I (0.1 #g), MDA5 (0.1 #g), or TLR3 (0.1 #g) expressing vectors weretransfected into HEK293 cells with the plasmid encoding the Riplet-DN fragment (0.1, 0.2, or 0.3 #g) in 24-well plates. After 24 h, the cells were stimulated with50 #g of poly(I-C) for 4 h, and relative luciferase activities were determined. F, Riplet-DN (100 ng) was co-transfected with full-length RIG-I (0, 50, 100, or 200 ng),RIG-I CARD-like region (dRIG-I) (0, 50, 100, or 200 ng), or C-terminal deleted RIG-I (RIG-I dRD) (0, 50, 100, or 200 ng) into HEK293 cells in 24-well plate, and reportergene assays were carried out.






IFN-$ promoter was only minimally activated by RIG-I inresponse to VSV (m.o.i.$ 1) during the early phase of infection(&12 h), the activitywas increased byRIG-I andRiplet (Fig. 3C).

Riplet was silenced by siRNA andthen VSV infected the cells. VSV-derived up-regulation of IFN-$mRNAwas started around 6 h post-infection, and Riplet siRNA signifi-cantly suppressed the increase ofIFN-$ mRNA at 6 h (Fig. 3D).Because VSV infection is mainlysensed by RIG-I, this is consistentwith the notion that Riplet pro-motes the RIG-I signaling. OtherIFN-inducible genes, IFIT1 andMX1, were expressed '8 h post-in-fection, and their expressions werealso suppressed by Riplet siRNA(Fig. 3, E and F).Riplet Exerts Protective Activity

against Viral Infection—Next weexamined the role of Riplet duringviral infection. Riplet and/or RIG-Iwere transiently expressed in thehuman cells by FuGENE HDreagents, and then the cells wereinfected with VSV or poliovirus (apositive-stranded RNA virus). Theviral titer of the supernatant wasdetermined 24 h post-infection.Under our conditions, expression ofRIG-I weakly inhibited VSV propa-gation. Co-expression of Riplet withRIG-I significantly suppressed VSVreplication especially at low m.o.i.,whereas Riplet alone did not sup-press VSV (Fig. 4, A and B, upperpanel). Therefore, a sufficientamount of RIG-I protein is requiredfor Riplet to exert antivirus activity.This requirement of RIG-I is alsoobserved in reporter gene analyses(Fig. 2). Under a similar setting, theantiviral effect of Riplet wasmargin-ally observed against poliovirus,which induces IFN-$ largely viaMDA5 (Fig. 4B, lower panel). Toassess the importance of endoge-nous Riplet for antivirus effect ofhuman cells, Riplet knockdown cellswere infected with viruses. In Ripletknockdown cells, the VSV titer wasconsistently increased comparedwith the control (p & 0.05) (Fig. 4C,left panel). In addition, infection ofRiplet knockdown cells with polio-virus resulted in only a slightincrease in the poliovirus titer com-

paredwith thecontrol (p'0.05) (Fig. 4C, rightpanel).Becausepolio-virus is mainly recognized by MDA5 but not RIG-I, this marginaleffect of Riplet onpoliovirus infectionwaswithin expectation (3, 28).

FIGURE 3. Knockdown analyses of Riplet. A, p125 luc reporter plasmid (0.1 #g), RIG-I expressing vector (0.1#g), and Riplet siRNA or control siRNA (10 pmol), which were purchased from Funakoshi Co. Ltd., were trans-fected into HEK293 cells in a 24-well plate with Lipofectamine 2000, and 48 h after transfection, the cells werestimulated with poly(I-C) for 6 h, and the cell lysate was prepared, and luciferase activities were measured.RT-PCR was carried out using total RNA extracted from cells 48 h after transfection. B, p125 luc reporter plasmid(0.1 #g), RIG-I expressing vector (0.1 #g), and siRNA, Riplet si-1, or control si-1 (10 pmol), which were purchasedfrom Applied Biosystems, were transfected into HEK293 cells with Lipofectamine 2000. 48 h after transfection,the cells were stimulated with poly(I-C) for 6 h. The cell lysate was prepared, and luciferase activities weremeasured. RT-PCR was carried out using total RNA extracted from cells 48 h after transfection. C, HEK293 cellswere transfected with the plasmids expressing RIG-I (0.1 #g) and/or Riplet (0.1 #g) with p125 luc reporterplasmid (0.1 #g) in 24-well plates. After 24 h, the cells were infected with VSV (m.o.i. $ 1) for 12 h. The luciferaseactivities of the cell lysates were measured. Expression of Riplet strongly enhanced IFN-$ promoter activationby VSV through RIG-I. D–F, siRNA (control si- or Riplet si-1) were transfected into HEK293 cells, and after 48 h,the cells were infected with VSV at m.o.i. $ 1. RNA was extracted at the indicated hours, and the quantitativePCR were carried out to detect the expression of IFN-$ (D), IFIT-1 (E), or Mx1 (F) mRNA. *, p & 0.05. GAPDH,glyceraldehyde-3-phosphate dehydrogenase.



Riplet and Riplet-DN Bind the Helicase and RD Regions ofRIG-I—Yeast two-hybrid analysis showed that a C-terminalregion of Riplet bound to the C-terminal region of RIG-I. Thiscytoplasmic interaction between Riplet and RIG-I was con-firmed by confocalmicroscopy inHeLa cells (supplemental Fig.S2). To further confirm the physical binding of Riplet to RIG-Iin human cells, we carried out immunoprecipitation analyses.Full-length Riplet was co-immunoprecipitated with RIG-I (Fig.5B), indicating that Riplet binds directly to RIG-I in humancells.To determine the region responsible for the RIG-I-Riplet

interaction, we constructed aRIG-I andRiplet deletion series asshown in Fig. 5A. Riplet-DN also bound to RIG-I (Fig. 5, B andC), indicating that the RING finger domain is dispensable forthe RIG-I-Riplet interaction. This is consistent with the notionthat the RING finger domain in ubiquitin ligase proteins isrequired for their interactions with ubiquitin-conjugatingenzymes (29). Unlike TRIM25, Riplet and Riplet-DN failed toco-precipitate the two CARD domains of RIG-I (dRIG-I) (Fig.5D). However, co-precipitation of the RIG-IC or RIG-RD frag-ments was observed (Fig. 5, E and F). RD-deleted RIG-I (RIG-IdRD) weakly associated with Riplet (Fig. 5G). Taken together,Riplet preferentially binds the RD and also weakly associateswith the helicase region of RIG-I with its C terminus. Reportergene analyses show that Riplet-DN only weakly suppressesRIG-I signaling and barely suppresses dRIG-I, which containsneither helicase nor RD region. Therefore, the physical interac-tion is correlated with the results of reporter activity.Riplet Promotes Ubiquitination of RIG-I—Because Riplet

shares 60% sequence similarity with TRIM25, we hypothesizedthat Riplet ubiquitinates RIG-I and that this modification leadsto activation of RIG-I signaling. To test this hypothesis, weexamined RIG-I ubiquitination. As expected, ubiquitination ofRIG-I was increased by co-expression of Riplet under two dif-ferent conditions (Fig. 6, A and B). The quantity of RIG-I ubiq-uitination was significantly high in the presence of Riplet (Fig.6C). RIG-I ubiquitination was suppressed if Riplet was replacedwith Riplet-DN (Fig. 6D and supplemental Fig. S4C). However,unlike TRIM25, Riplet binds to the C-terminal region of RIG-I.Therefore, we examined whether Riplet ubiquitinates theC-terminal region.We found that ubiquitination of RIG-ICwasenhanced by Riplet expression (Fig. 6E). Both RIG-I dRD andRIG-I RD were also ubiquitinated by expression of Riplet (Fig.6F; supplemental Fig. S4A and S5), suggesting that Riplet pro-motes ubiquitination of the helicase and RD domains of RIG-Iin a manner distinct from TRIM25.Ubiquitin is polymerized through its lysine residue. Lys-63-

linked polyubiquitination is frequently observed in signal trans-duction pathways (30). In contrast, Lys-48-linked polyubiquiti-nation usually leads to the degradation of protein through theproteasome. Indeed, TRIM25-mediated Lys-63-linked poly-ubiquitination activates the CARD-like region of RIG-I, andRNF125-mediated Lys-48-linked polyubiquitination leads tothe degradation of RIG-I (23, 25). We used K48R or K63Rmutated ubiquitin and found that K48R was incorporated nor-mally into RIG-IC, whereas polyubiquitination was decreasedby K63R (supplemental Fig. S4B). K63R mutation abolishedRIG-I RD polyubiquitination by Riplet (Fig. 6F). These data

FIGURE 4. Suppression of RNA viruses by Riplet. A, HEK293 cells weretransfected with RIG-I (0.1 #g) and/or Riplet (0.1 #g) expressing vectors.The total amount of transfected DNA (0.5 #g/well) in each well was keptconstant by adding empty vector (pEF-BOS). 24 h after transfection, thecells were infected with VSV at m.o.i. $ 0.1, and after 0, 18, or 24 h, CPE wasobserved by microscope. B, RIG-I (0.1 #g) and/or Riplet (0.1 #g) expressingplasmids were transfected to HEK293 cells in 24-well plates and incubatedfor 24 h. The total amount of transfected DNA (0.5 #g/well) in each wellwas kept constant by adding empty vector (pEF-BOS). The cells wereinfected with VSV (upper panel) or poliovirus (PV) (lower panel) at the indi-cated m.o.i. The viral titers in the culture media were measured 24 h afterinfection by plaque assay. Error bars represent standard deviation (n $ 3).*, p & 0.05. C, control or Riplet knockdown HEK293 cells were infected withVSV (left panel) or poliovirus (right panel) at m.o.i. $ 0.1. The viral titers inthe culture media were measured 26 h after infection by plaque assays.Knockdown of Riplet induced higher VSV titers compared with control(p & 0.05), but the increase observed in poliovirus-infected Riplet knock-down cells was not significant (p ' 0.05).








indicates that Riplet mediates Lys-63-linked polyubiquitination of theRIG-I C-terminal helicase and RDregion. Because Riplet-DN reducedthe RIG-I-mediated signaling, weexamined whether Riplet-DNreduced the RIG-I ubiquitination.As expected, Riplet-DN reducedRIG-I ubiquitination (Fig. 6D andsupplemental Fig. S4C). These ubiq-uitination assay data are consistentwith the notion that Riplet-medi-ated Lys-63-linked polyubiquitina-tion of RIG-I is required for full acti-vation of RIG-I signaling.We tried to determine the ubiq-

uitination sites of RIG-I using Lys-to-Ala (KA)-converting mutants.RIG-I has 25 Lys residues in itsC-terminal region. These Lys resi-dues of RIG-I were in turn mutatedto Ala, and the degree of ubiquitina-tion and IFN-$-inducing activitywere determined with each mutant.RIG-I-mediated IFN-$ promoteractivation was normally augmentedby co-expression of Riplet and 3KARIG-I. Co-expression of Riplet and5KA, however, and the ubiquitina-tion level of RIG-I and IFN-$-in-ducing activity were simultaneouslydecreased (Fig. 7, A and C). Riplet-dependent augmentation of IFN-$promoter activation was largelysuppressed when RIG-I was replacedwith 5KA RIG-I (Fig. 7B). Therefore,Lys-849 and Lys-851 of RIG-I werecrucial for RIG-I ubiquitination byRiplet. The results confirmed theimportance of ubiquitination of spe-cific Lys residues in the C-terminalregion of RIG-I and for RIG-I-medi-ated IFN-$ induction.

DISCUSSION

RIG-I plays a central role in therecognition of cytoplasmic viralRNA and is regulated by modifica-tion by small modifier ubiquitinor ubiquitin-like protein, ISG15.TRIM25 mediates Lys-63-linkedpolyubiquitination, which is essen-tial for RIG-I activation (23), andRNF125 mediates Lys-48-linkedpolyubiquitination (25). RIG-I alsoharbors ISG15 modification, al-though the role of ISG15 modifica-tion in vivo remains to be deter-

FIGURE 5. Physical interaction of Riplet with RIG-I. A, schematic representation of RIG-I or Riplet fragments usedfor immunoprecipitation analyses. B, HA-tagged Riplet (0.4 #g) or Riplet-DN (0.4 #g) were transfected intoHEK293FT cells in a 6-well plate with FLAG-tagged RIG-I (0.4 #g). HA-tagged Riplet or Riplet-DN were immunopre-cipitated (IP) with anti-HA antibodies, and samples were analyzed by Western blotting (WB) using an anti-FLAG oranti-HA antibody. The total amount of transfected DNA (2 #g/well) was kept constant by adding empty vector(pEF-BOS). C, HA-tagged Riplet-DN (0.4 #g) and FLAG-tagged RIG-I (0.4 #g) were transfected into HEK293FT cells ina 6-well plate. RIG-I was immunoprecipitated with anti-FLAG antibody, and samples were analyzed by Westernblotting using an anti-FLAG or -HA antibody. The total amount of transfected DNA (2#g/well) was kept constant byadding empty vector (pEF-BOS). D–F, interaction of HA-tagged Riplet or Riplet-DN with FLAG-tagged dRIG-I (D),RIG-IC (E), or RIG-I RD (F) was examined using immunoprecipitation assays. The proteins were expressed inHEK293FT cells, and HA-tagged Riplet was immunoprecipitated with anti-HA antibody, and samples were analyzedby Western blotting using an anti-FLAG or -HA antibody. G, FLAG-tagged RIG-I RD (0.4 #g) or RIG-I dRD (0.4 #g) wastransfected with HA-tagged Riplet (0.4 #g) into HEK293 FT cells in a 6-well plate, and 24 h after transfection, immu-noprecipitation was performed with anti-FLAG antibody and analyzed by Western blotting. The total amount oftransfected DNA (2 #g/well) was kept constant by adding empty vector (pEF-BOS). WCE, whole cell extract.




mined (21, 22, 31). Although Ripletand TRIM25 share 60% sequencesimilarity, the ubiquitination ofRIG-I by Riplet is distinct from thatby TRIM25; Riplet ubiquitinates theC-terminal region of RIG-I, whereasTRIM25 ubiquitinates its CARD-like region. These findings are alsosupported by the fact that neitherRiplet nor Riplet-DN promoted orinhibited the activation of the IFN-$promoter by expression of theRIG-ICARD-like region (data not shown).It has been reported that ubiquitina-tion of the CARD-like region ofRIG-I by TRIM25 is critical for RIG-I-IPS-1 signaling (23). However,how this CARD ubiquitination isessential for activation of IPS-1 byRIG-I remains undetermined. Herewe emphasize the importance ofRIG-I C-terminal ubiquitination forIFN-$ induction and the antiviralresponse. Because the C-terminalRD region inhibits the IFN inducingactivity of the CARD-like region ofRIG-I, it is reasonable that RIG-IC-terminal ubiquitination by Ripletinhibits the conversion from theactive to inactive form of RIG-I pro-tein after binding to viral RNA. Thisinitial stabilization of RIG-I viaubiquitination by Riplet would pro-vide a sufficient structure for RIG-Ito maintain the accessibility toTRIM25 and facilitate TRIM25-me-diated ubiquitination of the CARD-like region of RIG-I, whichmay leadto potential activation of IPS-1.RIG-I is an IFN-inducible RNA

helicase that is expressed at ex-tremely low levels in resting cells (6).Initial penetration of viruses allowsgeneration of 5!-triphosphate RNAand/or double strand RNA followedby induction of IFN-$ production.This early response to viral infec-tions triggers up-regulation of RIG-I/MDA5 and TLR3, leading torobust IFN-$ production (3, 32, 33).We favor the interpretation of ourpresent findings that during theearly stages of viral infection withtrace amounts of RIG-I and viralRNAs, Riplet helps host cells rear-range RIG-I conformation to acti-vate IPS-1. This issue will need fur-ther proof because it is difficult to

FIGURE 6. Ubiquitination of RIG-I by Riplet. A and B, FLAG-tagged RIG-I (0.4 #g), Riplet (0.4 #g), andHA-tagged ubiquitin (0.4 #g) expressing vectors were transfected into HEK293FT cells in 6-well plates. Thetotal amount of transfected DNA (2 #g/well) was kept constant by adding empty vector (pEF-BOS). FLAG-tagged RIG-I was immunoprecipitated (IP) using an anti-FLAG antibody, and washed with the buffercontaining 150 mM NaCl (A) or 1 M NaCl (B). The immunoprecipitates were separated with 8% acrylamidegel and analyzed by Western blotting (WB) using antibodies against HA tag (ubiquitin) or FLAG (RIG-I).Riplet was co-immunoprecipitated with FLAG-tagged RIG-I in A but could not co-immunoprecipitate in Bbecause of high salt condition. Expression of Riplet enhanced the ubiquitination of RIG-I. Different gelconditions were employed in A and B. C, ubiquitinated RIG-I was quantitated with NIH image software. **,p & 0.01. D, FLAG-tagged RIG-I (0.4 #g) was transfected into HEK293 FT cells in a 6-well plate withHA-tagged Riplet (0.4 #g) or Riplet-DN (0.4 #g) and HA-tagged ubiquitin, and immunoprecipitation wascarried out with anti-FLAG antibody. The total amount of transfected DNA (2 #g/well) was kept constantby adding empty vector (pEF-BOS). The samples were analyzed with 10% acrylamide gel to clearly sepa-rate Riplet from Riplet-DN and stained by Western blotting. E, ubiquitination of RIG-IC was also promotedby Riplet expression. HEK293FT cells were transfected with the plasmids encoding RIG-IC (0.4 #g), Riplet(0.4 #g), and/or HA-tagged ubiquitin (0.4 #g) in a 6-well plate, and 24 h after transfection, cell lysates wereprepared. The total amount of transfected DNA (2 #g/well) was kept constant by adding empty vector(pEF-BOS). FLAG-tagged RIG-ICs were immunoprecipitated with anti-FLAG antibodies, and the proteinswere analyzed by Western blotting. F, Ub-K63R are HA-tagged ubiquitin in which the lysine 3 residueswere substituted with arginine. The HA-tagged Ub-K63 expressing vectors (1.2 #g), FLAG-tagged RIG-IC(0.4 #g), and/or Riplet (0.4 #g) were transfected into HEK293FT cells in 6-well plates and analyzed asshown in A–D. The total amount of transfected DNA (2 #g/well) was kept constant by adding empty vector(pEF-BOS). Ub-K63R was not incorporated into polyubiquitin chain of RIG-I RD. WCE, whole cell extract.



visualize RNRs and viral RNAs in the early infection stage andto understand themechanisms that allow viruses to uncoat intonaked viral RNA and to replicate.We have provided several lines of evidence indicating that

Riplet complements RIG-I-mediated IFN-$ induction uponviral infection by both Riplet siRNA and overexpression analy-ses. The C-terminal lysines (849 and 851) of RIG-I are criticalfor Riplet-mediated RIG-I ubiquitination. However, our dataindicate that Riplet alone was unable to induce IFN-$ produc-tion and essentially required RIG-I to confer IFN-$ induction.Furthermore, Riplet is not ubiquitously distributed over the

organs tested. Ubiquitination ofRIG-I induced by poly(I-C) orviruses was accelerated in cellspre-transfected with Riplet. Hence,Riplet works case-sensitive to up-regulate RIG-I antiviral activity pre-dominantly in some organs. Thephysiological meaning of this re-sponse will be clarified by knock-out study.Unexpectedly, the siRNA experi-

ments were not robust with regardto VSV replication. Possible expla-nations for this are as follows: 1) thedegree of gene silencing is not soprofound that the proteins remainin the cells; 2) there are a number ofvirus-mediated IFN-inducing path-ways capable of compensating eachother, so that disruption of one fac-tor does not cause a profound effecton VSV replication. Furthermore,in VSV-infected Riplet-knockdowncells, IFN-$ levels were reducedeven at m.o.i. $ 1 (Fig. 3D), andaccordingly, virus susceptibility wasincreased at m.o.i. $ 0.1 (Fig. 4C),whereas in Riplet-overexpressingcells, antiviral activity was observedonly at low m.o.i. (Fig. 4B). We useddifferent transfection reagents andcell conditions in the knockdownand overexpression experiments toobtain high transfection efficiencyin each. These conditional differ-ences in knockdown and overex-pression analyses might cause partof the discrepancy between the tworesults on Riplet antiviral activity.Another possibility to explain theapparent inconsistencies betweenoverexpression and knockdownanalyses is that high amounts ofRiplet efficiently activate the RIG-Isignaling, but low amounts areinsufficient for RIG-I activation inhigh m.o.i.-infecting human cells.

High amounts of Riplet with overexpressed RIG-I would conferthe ability on cells to respond to very low amounts of VSV asobserved in the lowm.o.i. experiments. Again, riplet knock-outmicewould revealwhether it is absolutely required for potentialRIG-I activation.How viral RNAs select RIG-I rather than dicers or the trans-

lationmachinery is also unknown. During natural infection it islikely that the number of the initial invading virions would be atmost several copies/cell. Uncoated viral RNA may assemble acomplex consisting of viral and host molecules required forreplication. We assume that cells are equipped with various

FIGURE 7. The C-terminal two lysine residues of RIG-I are important for ubiquitination by Riplet. A, RIG-IC-terminal lysine residues were substituted with alanine. RIG-I 3KA mutant protein harbors the triple muta-tions, K888A, K907A, and K909A. The five lysine residues, Lys-849, Lys-851, Lys-888, Lys-907, and Lys-909, werereplaced with alanine in RIG-I 5KA mutant. The plasmid carrying wild-type (100 ng/well), RIG-I 3KA (100ng/well), RIG-I 5KA (100 ng), or Riplet (100 ng) were transfected into HEK293 cells in a 24-well plate togetherwith p125 luc reporter plasmid (100 ng/well). The amount of transfected DNA was kept constant by addingempty vector. After 24 h, the luciferase activities were measured. B, wild-type RIG-I (100 ng), RIG-I 5KA mutant(100 ng), or empty vector (100 ng) was transfected into HEK293 cells in a 24-well plate together with p125 lucreporter plasmids and HCV 3!-untranslated region poly(U/UC) RNA (25 ng), which is synthesized in in vitrotranscription by T7 RNA polymerase. The amount of transfected DNA was kept constant by adding emptyvector. 24 h after transfection, luciferase activities were measured. RIG-I 5KA mutant hardly responded topoly(U/UC) RNA. C, to observe the ubiquitinated RIG-I more clearly, we used 800 ng/well of Riplet and HA-Ubexpression vector for the following transfection. HEK293FT cells in a 6-well plate were transfected with theplasmids encoding RIG-I (400 ng/well), RIG-I 5KA (400 ng/well), Riplet (800 ng/well), and/or HA-Ub (800ng/well). The total amount of DNA was kept constant by adding the empty vector. 24 h after the transfection,the cell lysates were prepared, and the immunoprecipitation was carried out using anti-FLAG antibodies. Theimmunoprecipitates were analyzed by Western blotting with anti-HA or FLAG antibodies.



molecular arms to sensitively detect viral RNA. The molecularcomplexes sensing viral RNAmay not be so simple that we willbe able to identify more molecules than Riplet as enhancers forintegral RNA recognition. In either case, yeast screening will bea good strategy to pick up such proteins in other RNA recogni-tion systems. A molecular switch selecting IFN induction byvirus RNA will then be clarified.We show that the ubiquitination sites targeted by Riplet are

the helicase and RD domains of RIG-I but not its CARD-likedomains in contrast to TRIM25. Riplet may be a complementfactor of the reported TRIM25 function for RIG-I activation(23). A previous report (25) failed to polyubiquitinate the RIG-Iprotein by TRIM25 alone. If Riplet were added to TRIM25 forRIG-I ubiquitination in the previous study, Riplet would haveenabled TRIM25 to polyubiquitinate the RIG-I CARD-likeregion. Further studies using TRIM25 and Riplet will berequired to clarify this point.Based on our results, we propose that RIG-I-like receptors

form a molecular complex that efficiently recognizes low copynumbers of viral RNA. Riplet is implicated in the RIG-I com-plex to enhance viral RNA response in some organs. In thiscontext, MDA5-associated molecules might also exist in thecytoplasm to augment IFN output. Although MDA5 possessesthe RD domain, it fails to recruit Riplet (data not shown) oraugment IFN-$-induction in conjunction with Riplet (Fig. 2E).Because RLR-associated molecules naturally reside in cells andfacilitate inhibition of low dose viral infection until RLRsbecome expressed, theymay be useful therapeutic targets for anearly phase antiviral immunotherapy.

Acknowledgments—We thankDr.M. Sasai in our laboratory for tech-nical instructions for assay of RIG-I functions andDrs. K. Shimotohno(Keio University), T. Taniguchi (University of Tokyo), and T. Fujita(Kyoto University) for their critical discussions.

REFERENCES1. Takeuchi, O., and Akira, S. (2008) Curr. Opin. Immunol. 20, 17–222. Honda, K., Takaoka, A., and Taniguchi, T. (2006) Immunity 25, 349–3603. Kato, H., Takeuchi, O., Sato, S., Yoneyama,M., Yamamoto,M.,Matsui, K.,

Uematsu, S., Jung, A., Kawai, T., Ishii, K. J., Yamaguchi, O., Otsu, K.,Tsujimura, T., Koh, C. S., Reis e Sousa, C., Matsuura, Y., Fujita, T., andAkira, S. (2006) Nature 441, 101–105

4. Venkataraman, T., Valdes, M., Elsby, R., Kakuta, S., Caceres, G., Saijo, S.,Iwakura, Y., and Barber, G. N. (2007) J. Immunol. 178, 6444–6455

5. Yoneyama, M., Kikuchi, M., Matsumoto, K., Imaizumi, T., Miyagishi, M.,Taira, K., Foy, E., Loo, Y. M., Gale, M., Jr., Akira, S., Yonehara, S., Kato, A.,and Fujita, T. (2005) J. Immunol. 175, 2851–2858

6. Yoneyama, M., Kikuchi, M., Natsukawa, T., Shinobu, N., Imaizumi, T.,Miyagishi, M., Taira, K., Akira, S., and Fujita, T. (2004) Nat. Immunol. 5,730–737

7. Hornung, V., Ellegast, J., Kim, S., Brzozka, K., Jung, A., Kato, H., Poeck, H.,Akira, S., Conzelmann, K. K., Schlee, M., Endres, S., and Hartmann, G.(2006) Science 314, 994–997

8. Pichlmair, A., Schulz, O., Tan, C. P., Naslund, T. I., Liljestrom, P., Weber,F., and Reis e Sousa, C. (2006) Science 314, 997–1001

9. Saito, T., Hirai, R., Loo, Y.M., Owen, D., Johnson, C. L., Sinha, S. C., Akira,S., Fujita, T., and Gale, M., Jr. (2007) Proc. Natl. Acad. Sci. U. S. A. 104,582–587

10. Kawai, T., Takahashi, K., Sato, S., Coban, C., Kumar, H., Kato, H., Ishii,K. J., Takeuchi, O., and Akira, S. (2005) Nat. Immunol. 6, 981–988

11. Meylan, E., Curran, J., Hofmann, K., Moradpour, D., Binder, M., Barten-schlager, R., and Tschopp, J. (2005) Nature 437, 1167–1172

12. Seth, R. B., Sun, L., Ea, C. K., and Chen, Z. J. (2005) Cell 122, 669–68213. Xu, L. G., Wang, Y. Y., Han, K. J., Li, L. Y., Zhai, Z., and Shu, H. B. (2005)

Mol. Cell 19, 727–74014. Rothenfusser, S., Goutagny, N., DiPerna, G., Gong, M., Monks, B. G.,

Schoenemeyer, A., Yamamoto, M., Akira, S., and Fitzgerald, K. A. (2005)J. Immunol. 175, 5260–5268

15. Loo, Y. M., Fornek, J., Crochet, N., Bajwa, G., Perwitasari, O., Martinez-Sobrido, L., Akira, S., Gill, M. A., Garcia-Sastre, A., Katze,M. G., andGale,M., Jr. (2008) J. Virol. 82, 335–345

16. Komuro, A., and Horvath, C. M. (2006) J. Virol. 80, 12332–1234217. McWhirter, S. M., Tenoever, B. R., and Maniatis, T. (2005) Cell 122,

645–64718. Saha, S. K., Pietras, E. M., He, J. Q., Kang, J. R., Liu, S. Y., Oganesyan, G.,

Shahangian, A., Zarnegar, B., Shiba, T. L., Wang, Y., and Cheng, G. (2006)EMBO J. 25, 3257–3263

19. Kayagaki, N., Phung, Q., Chan, S., Chaudhari, R., Quan, C., O’Rourke,K.M., Eby,M., Pietras, E., Cheng,G., Bazan, J. F., Zhang, Z., Arnott, D., andDixit, V. M. (2007) Science 318, 1628–1632

20. Lin, R., Yang, L., Nakhaei, P., Sun, Q., Sharif-Askari, E., Julkunen, I., andHiscott, J. (2006) J. Biol. Chem. 281, 2095–2103

21. Zhao, C., Denison, C., Huibregtse, J. M., Gygi, S., and Krug, R. M. (2005)Proc. Natl. Acad. Sci. U. S. A. 102, 10200–10205

22. Arimoto, K., Konishi, H., and Shimotohno, K. (2008) Mol. Immunol. 45,1078–1084

23. Gack, M. U., Shin, Y. C., Joo, C. H., Urano, T., Liang, C., Sun, L., Takeuchi,O., Akira, S., Chen, Z., Inoue, S., and Jung, J. U. (2007) Nature 446,916–920

24. Urano, T., Saito, T., Tsukui, T., Fujita, M., Hosoi, T., Muramatsu, M.,Ouchi, Y., and Inoue, S. (2002) Nature 417, 871–875

25. Arimoto, K., Takahashi, H., Hishiki, T., Konishi, H., Fujita, T., and Shimo-tohno, K. (2007) Proc. Natl. Acad. Sci. U. S. A. 104, 7500–7505

26. Sasai, M., Shingai, M., Funami, K., Yoneyama, M., Fujita, T., Matsumoto,M., and Seya, T. (2006) J. Immunol. 177, 8676–8683

27. Douglas, J., Cilliers, D., Coleman, K., Tatton-Brown, K., Barker, K., Bern-hard, B., Burn, J., Huson, S., Josifova, D., Lacombe, D.,Malik,M.,Mansour,S., Reid, E., Cormier-Daire, V., Cole, T., and Rahman, N. (2007) Nat.Genet. 39, 963–965

28. Barral, P.M.,Morrison, J.M., Drahos, J., Gupta, P., Sarkar, D., Fisher, P. B.,and Racaniello, V. R. (2007) J. Virol. 81, 3677–3684

29. Seol, J. H., Feldman, R. M., Zachariae, W., Shevchenko, A., Correll, C. C.,Lyapina, S., Chi, Y., Galova, M., Claypool, J., Sandmeyer, S., Nasmyth, K.,Deshaies, R. J., Shevchenko, A., and Deshaies, R. J. (1999) Genes Dev. 13,1614–1626

30. Pickart, C. M. (2004) Cell 116, 181–19031. Kim, M. J., Hwang, S. Y., Imaizumi, T., and Yoo, J. Y. (2008) J. Virol. 82,

1474–148332. Alexopoulou, L., Holt, A. C., Medzhitov, R., and Flavell, R. A. (2001) Na-

ture 413, 732–73833. Tanabe, M., Kurita-Taniguchi, M., Takeuchi, K., Takeda, M., Ayata, M.,

Ogura, H., Matsumoto, M., and Seya, T. (2003) Biochem. Biophys. Res.Commun. 311, 39–48

34. Yoneyama, M., Suhara, W., Fukuhara, Y., Fukuda, M., Nishida, E., andFujita, T. (1998) EMBO J. 17, 1087–1095



Documents

PLK1 Down-Regulates Parainfluenza Virus 5 Gene Expression