Virology 388 (2009) 221–229

Contents lists available at ScienceDirect

Virology

j ourna l homepage: www.e lsev ie r.com/ locate /yv i ro

Rescue of internal initiation of translation by RNA complementation providesevidence for a distribution of functions between individual IRES domains

Paula Serrano, Jorge Ramajo, Encarnación Martínez-Salas ⁎Centro de Biología Molecular Severo Ochoa, Consejo Superior de Investigaciones Científicas-Universidad Autónoma de Madrid, Cantoblanco 28049 Madrid, Spain

⁎ Corresponding author. Centro de Biologia MolecuNicolas Cabrera 1, Cantoblanco, 28049, Madrid, Spain. F

E-mail address: emartinez@cbm.uam.es (E. Martínez

0042-6822/$ – see front matter © 2009 Elsevier Inc. Adoi:10.1016/j.virol.2009.03.021

a b s t r a c t

a r t i c l e i n f o

Article history:Received 9 February 2009Returned to author for revision6 March 2009Accepted 23 March 2009Available online 21 April 2009

Keywords:PicornavirusIRESRNA–RNA interactionsTrans-complementationRNA structureIRES–protein interactioneIF4GProtein synthesisFMDV infection

Picornavirus RNAs initiate translation using an internal ribosome entry site (IRES)-dependent mechanism.The IRES element of foot-and-mouth disease virus (FMDV) is organized in domains, being different fromeach other in RNA structure and RNA–protein interaction. Wild-type transcripts provided in trans rescuedefective FMDV IRES mutants. Complementation, however, was partial since translation efficiency of themutant RNAs was up to 10% of the wild type IRES. We report here that mutations diminishing the RNA–RNAinteraction capacity induced a decrease in IRES rescue. On the other hand, IRES transcripts bearing mutationsthat reorganize the RNA structure of the apical region of central domain, although weakly, complementdefective IRES that are unable to interact with the initiation factor eIF4G in a separate domain. Together,these results suggest that IRES rescue may involve RNA-mediated contacts between defective elements, eachcarrying a defect in a separate domain but having the complementing one with the appropriate structuralorientation and/or ribonucleoprotein composition. Our observations further support the essential role of thecentral domain of the FMDV IRES during protein synthesis and demonstrate that there is a division offunctions between the IRES domains.

© 2009 Elsevier Inc. All rights reserved.

Introduction

The Picornaviridae family includes a group of pathogens that arethe causative agents of important diseases worldwide. Foot-and-mouth disease virus (FMDV), the prototypic member of theaphthovirus genus of this family, is the etiological agent of adevastating disease of livestock (Sobrino et al., 2001). The viralparticle is composed of a protein capsid that contains an infectiouspositive-sense RNA genome. The viral RNA encodes a singlepolyprotein, which is processed in infected cells yielding variouspolypeptide precursors and the mature viral proteins (Belsham andMartinez-Salas, 2004).

Noncoding regions flanking both ends of the viral RNA containspecific structural elements involved in the control of translation andreplication of the FMDV genome (Fernandez-Miragall et al., 2009; Saizet al., 2001). Translation initiation of all picornavirus RNAs iscontrolled by an internal ribosome entry site (IRES) element locatedin the long 5′ untranslated region of the viral RNA (Belsham, 2009;Martinez-Salas et al., 2008; Semler and Waterman, 2008). Translationof the viral genome is the first intracellular step of the picornavirusinfection cycle, and thus, the effectiveness of infection is dependent onthe correct function of the IRES. Therefore, the IRES region constitutes

lar Severo Ochoa, CSIC-UAM,ax: +34 911964420.-Salas).

ll rights reserved.

a target for antiviral drugs aimed at blocking the virus life cycle(Gutierrez et al., 1994; Vagnozzi et al., 2007) as well as being adeterminant of viral pathogenesis.

The FMDV IRES encompasses about 450 nt organized in domains,termed 1 to 5 (Fig. 1). For activity this element requires the translationinitiation factors eIF4G, eIF4A, eIF3, eIF2 and other IRES-transactingfactors (Andreev et al., 2007, Martinez-Salas et al., 2008). However, ithas been shown that the region responsible for interaction with eIFsalone does not constitute an IRES element (Fernandez-Miragall et al.,2009). In addition, FMDV IRES function depends on its structuralorganization (Fernandez-Miragall and Martinez-Salas, 2003; Marti-nez-Salas, 2008). These observations strongly suggest that the entireIRES region works as a single entity to achieve internal initiation.

The distal domains within the FMDV IRES structure are involvedin interactions with host factors including, among other proteins, thetranslation initiation factor eIF4G (Lopez de Quinto et al., 2001;Lopez de Quinto and Martinez-Salas, 2000; Pacheco et al., 2009;Pilipenko et al., 2000; Stassinopoulos and Belsham, 2001). In supportof the relevance of the interaction of eIF4G with the FMDV IRES,substitutions of nucleotides 309–313 and 411–412 which disrupt thestem structure of domain 4 or alter the sequence of the A-bulge inthis IRES region (Fig. 1) impair eIF4G binding and, concomitantly,abolish IRES activity (Lopez de Quinto and Martinez-Salas, 2000).

The role performed by the central domain of the FMDV IRES thatoccupies a significant portion of its entire sequence still remainselusive. Just a few interactions with host factors have been identified

Fig. 1. RNA structure organization of the FMDV IRES. RNA structure is depicted according to Fernandez-Miragall et al. (2009). Nucleotides encompassing the IRES domains referred toin the text, as well as the conserved motifs and the binding site of eIF4G, are indicated.

222 P. Serrano et al. / Virology 388 (2009) 221–229

by riboproteomic analysis in this region in comparison to the largenumber of factors identified as binding to the 3′end region of theIRES (Pacheco et al., 2008). A distinctive feature of the centraldomain (termed 3, Fig. 1) is a cruciform structure at its tip(Fernandez-Miragall and Martinez-Salas, 2003). This region containstwo conserved motifs, GNRA and RAAA (N stands for any nucleotide;and R, purine) as defined by mutational analysis, that do not toleratenucleotide substitutions (Lopez de Quinto and Martinez-Salas, 1997).A single nucleotide substitution of GUAA to GUAG in the GNRA motifinduces about 90% reduction of IRES activity. Furthermore, substitu-tions of GUAA to UGCG, replacing the GNRA motif by a UNCGtetraloop, were not compatible with IRES activity either (Fernandez-Miragall and Martinez-Salas, 2003). A similar reduction of activity isinduced by nucleotide substitutions of GAAAA to CGCCC in the RAAAmotif. The proximal region of this domain termed stem 3 is orga-nized as a base-paired structure interrupted with bulges (Fig. 1). Thisregion includes several non-canonical base pairings (Fernandez-Miragall et al., 2009) and a helical structure that is specificallyrequired for IRES activity (Martinez-Salas et al., 1996). The centraldomain has been suggested to play a regulatory role during internalinitiation, in which the proximal region serves as a flagpole thatholds the apical region in a conformation appropriate for itsrecognition by the translation machinery (Martinez-Salas, 2008).

We have previously shown the capacity of the FMDV IRES domainsto form strand-specific RNA–RNA interactions in vitro (Ramos andMartinez-Salas, 1999). The efficiency of these interactions wasdependent on the RNA concentration, the presence of divalent ions

and the temperature of binding. In addition, RNA–RNA interactionswere dependent on the specific IRES region encompassed by thetranscript used as probe. For instance, the hairpin that contains theGNRAmotif was able to form RNA complexes only with domain 3, butit did not interact with the distal domains (Fernandez-Miragall et al.,2006). The formation of RNA–RNA interactions dependent on domain3 of the FMDV IRES is likely due to two types of motifs. The first oneconcerns tertiary interactions mediated by the conserved GNRA motiflocated in the apical region of the central domain (Fernandez-Miragallet al., 2009). The second one concerns the stem structure located atthe proximal region of this domain, which self-interacts veryefficiently (Ramos and Martinez-Salas, 1999).

The ability of the FMDV IRES element to generate RNA–RNAcomplexes in vitro suggested potential intermolecular RNA interac-tions with other viral RNAs present in the cell cytoplasm, which couldlead to trans-complementation effects, and thus, act to enhance theactivity of defective IRES variants. In order to test this possibility, weused various constructs expressing FMDV IRES transcripts to rescuedefective IRES elements in transfected cells. Picornavirus IRESelements are amenable to trans-complementation (Drew and Bel-sham, 1994; Stone et al., 1993; Van Der Velden et al., 1995). Previousstudies using defective encephalomyocarditis virus (EMCV) IRESshowed that this event requires the expression of the positive senseRNA and that it is independent of RNA recombination (Roberts andBelsham, 1997).

Complementation in trans is typically applied to the study ofprotein function. However, it may also help to underscore nucleic acid

Fig. 2. (A) Schematic representation of transcripts expressed from plasmids used in this study. The complementing RNAs, expressed from pGEM vectors under the control of the T7promoter, are depicted on the top panel. Black bars are used for FMDV IRES transcripts and white bar for HCV IRES. FMDV IRES transcripts carrying mutations GUAG181 and E2117–120are described in the text. The bottom panel depicts four defective bicistronic constructs (4G-binding312–313, IRES GUAG181, and IRES CGCCC199–203) encoding RNA of the form CAT-IRES-luciferase used in the rescue assay. See Fig. 1 for nucleotide numbering and distribution of domains 1–2, 3, stem 3 and 4–5. (B) Rescue of IRES activity as a function of theconcentration of the plasmid expressing the complementing RNA. Different amounts of the plasmid expressing exclusively the wild type IRES transcript, ranging between 0 and100 ng/2×104 cells were cotransfected with the bicistronic construct FMDV IRES 4G-binding312–313, which expresses a mutant RNA that is defective in eIF4G binding (squares)(Lopez de Quinto and Martinez-Salas, 2000). IRES activity rescue was calculated as the % of luciferase/CAT activity, normalized to the value obtained for the bicistronic constructcarrying the wild type IRES, transfected in parallel. The control rescue assay was conducted using pGEM3 empty vector to provide same amount of plasmid DNA in the transfectionmixture (circles). Assays were done always in triplicate M24wells, performed at least in two independent experiments. Error bars correspond to the standard deviation of the mean.

223P. Serrano et al. / Virology 388 (2009) 221–229

functions. Knowledge of the structural organization of the FMDV IRESas well as the involvement of the discrete IRES domains in RNA–protein interaction has received much attention in recent years(Fernandez-Miragall et al., 2009; Pacheco et al., 2008), providing therational basis for interpretation of complementation studies. We showhere that not only the wild type IRES sequence but also defective IRESvariants were able to rescue, although weakly, other defective IRESelements affected in the interaction with eIF4G. However, mutantswith a diminished RNA–RNA interaction capacity exhibited adecreased efficiency to rescue IRES activity. Thus, IRES complementa-tion reveals the essential role played by individual domains during

internal initiation of translation, and demonstrates a distribution offunctions performed by each structural domain.

Results

Internal initiation defects owing to lack of eIF4G interaction are rescuedby IRES transcripts expressed in trans

We have used various constructs expressing the FMDV IRES todetermine their capacity to rescue the activity of severely defectiveIRES variants driving translation of a reporter cistron in transfected

Fig. 3. Rescue of IRES activity by the full length FMDV IRES transcript. (A) RNA structure reorganization of GUAG or CGCCC IRES mutants, according to RNA probing analysis(Fernandez-Miragall and Martinez-Salas, 2003, Fernandez-Miragall et al., 2006). (B) Bicistronic plasmids (25 ng/2×104 cells), carrying defects in RNA structure of domain 3 (GUAGand CGCCC) or in eIF4G binding, were cotransfected with the plasmid expressing the wild type IRES sequence (12.5 ng/2×104 cells). IRES activity rescue was calculated as in Fig. 2B.The empty vector, pGEM3, was used in the negative control to supplement the amount of DNA in the transfection (white bars) and the plasmid expressing the HCV IRES was used totest the sequence specificity required for complementation (gray bars).

224 P. Serrano et al. / Virology 388 (2009) 221–229

cells (Fig. 2A). To this end, BHK-21 cells were transfected with twoplasmids, one expressing the defective IRES within a bicistronic RNAand the second one, expressing solely the complementing IREStranscript. The empty vector pGEM3 was used to supplement thedefective IRES construct when necessary to provide equal amounts ofplasmid DNA being present in the transfection mixture. IRES activitywas quantified as the expression of luciferase (IRES-dependent)normalized to that of chloramphenicol acetyl transferase (CAT) (cap-dependent) and then, made relative to the value obtained for the wildtype bicistronic construct, cotransfected with the empty vector. First, aconcentration range of the complementing plasmid was tested to setup the optimal conditions to rescue IRES activity of a severelydefective IRES mutant bearing nucleotide substitutions A312–A313 toGC (Fig. 1), which abolished binding to eIF4G and inhibited IRESactivity (Lopez de Quinto and Martinez-Salas, 2000). Relative to theefficiency of protein synthesis displayed by the mutant bicistronicconstruct cotransfected with the empty vector (0.14% of the wild typeactivity), translation activity observed following coexpression of thewild type transcript increased up to 50-fold, reaching levels that wereabout 10% of the wild type IRES values (Fig. 2B). This result indicatedthat this process, although partial, reconstitutes the IRES function.

It is worth noting that none of the plasmids expressing the variouscomplementing RNAs (the full length IRES sequence or its individualdomains, 2, 3, 4 in either, wild type or mutated version) usedthroughout this report interfered with the activity of the wild typeFMDV IRES (data not shown). Therefore, the conditions used tomeasure IRES rescue are optimal to detect complementation, withoutside effects due to sequestration of factors by overexpression of thecomplementing RNA.

RNA structure defects disrupting distant interactions in domain 3 can becomplemented in trans by the wild type IRES transcript

As aforementioned, specific RNA structural motifs that mediatethe local structure of the central domain of the FMDV IRES areresponsible for the efficiency of internal initiation (Fernandez-Miragall and Martinez-Salas, 2003). To asses whether defectsaltering the RNA structural organization of the FMDV IRES can berescued in trans, we measured the ability of a wild type IRES tocomplement the activity of defective IRES variants carrying nucleo-tide substitutions in conserved motifs of the apical region of domain3. Thus, we made use of mutations affecting the GNRA motif (GUAA

Fig. 4. Rescue of IRES activity by transcripts encompassing the central domain of theFMDV IRES alone. Defective FMDV IRES expressed from bicistronic plasmids (25 ng/2×104 cells) carrying substitutions in the GNRA motif (GUAG, UGCG), or the RAAAmotif (CGCCC) were cotransfected with the plasmid (12.5 ng/2×104 cells) expressingthe wild type IRES sequence (black bars), the domain 3 (gray bars), the stem 3 region(striped bars), or the domains 1–2 (dotted bars). IRES activity rescue was calculated asin Fig. 2B.

225P. Serrano et al. / Virology 388 (2009) 221–229

in FMDV C-S8) and the RAAA motif (AAAA in FMDV C-S8) thatimpair IRES activity reducing it to 0.5–1% of the wild type activity(Lopez de Quinto and Martinez-Salas, 1997) concomitant with a localRNA structure reorganization of this region (Fig. 3A), as determinedby RNA probing (Fernandez-Miragall and Martinez-Salas, 2003).Thus, the IRES mutant GUAG (carrying a single substitution, A181 to Gwithin the GNRA motif), within a bicistronic construct wascotransfected with a plasmid expressing solely the wild type FMDVIRES sequence. The empty vector, pGEM3, was used to supplementthe amount of DNA in the transfection control without anycomplementing IRES construct. As shown in Fig. 3B, a significantincrease (N10-fold) in IRES activity was observed relative to thatmeasured with the empty vector. A similar result was obtained withthe mutant bearing substitutions in A199AAAG203 to CGCCC withinthe RAAA conserved motif (Fig. 3B), indicating that RNA structuredefects in the apical region of domain 3 can be complemented intrans by the wild type IRES transcript.

Under these conditions, lack of IRES activity owing to defects ininteraction with the translation initiation factor eIF4G in a mutant ofdomain 4 carrying a double substitution, A312–A313 to GC) (Lopez deQuinto and Martinez-Salas, 2000) was also amenable to trans-complementation by the wild type IRES transcript (Figs. 2B and 3B).

To determine the specificity of the complementation, we carriedout a parallel assay with the hepatitis C virus (HCV) IRES, an RNA thatis also efficient in driving internal initiation of translation under thesame conditions used for FMDV IRES (Saiz et al., 1999), but it differsfrom the latter in primary sequence, secondary RNA structure andrequirement of initiation factors (Lukavsky, 2009; Martinez-Salaset al., 2008). Relative to the values obtained with the empty vector, noincrease in IRES activity was observed either in the GUAG, the CGCCCor the eIF4G-binding defective IRES elements (Fig. 3B). Therefore, ahomologous IRES sequence is required to rescue the activity of thedefective element. This result is in full agreement with the lack ofEMCV IRES complementation by other picornavirus IRES elements(Roberts and Belsham, 1997).

The wild type domain 3 transcript is sufficient to complement RNAstructure defects residing within this IRES region

Next, to test if the entire IRES sequence was needed in trans toexert positive complementation effects, we used a transcript expres-sing solely the sequences corresponding to the central domain(D3wt). In comparison to the empty vector, 4 to 10-fold stimulationof IRES activity was observed in two different bicistronic constructscarrying mutations in the GNRA motif, GUAG and UGCG (Fig. 4). Therescued activity of these defective elements reached values of 2.7 and8.3% of the wild type IRES, respectively. As above, similar results wereobserved with the CGCCC defective IRES mutant. This result demon-strated that the wild type domain 3 transcript is sufficient tocomplement defects arising within this region of the IRES.

In contrast, the cotransfection of a transcript termed stem 3, cor-responding exclusively to the proximal region of this domain (Figs. 1and 2A), did not modify the background IRES activity observed in eachdefective mutant (Fig. 4). Similarly, co-expression of a transcript cor-responding to domains 1–2 (D 1–2) failed to rescue the defective IRES.Thus, in order to rescue IRES activity, the complementing RNA shouldprovide in trans the entire domain corresponding to themutated one inthe defective IRES, with the proper structural organization.

Defects in RNA–RNA interaction within domain 3 can affect the efficiencyof IRES trans-complementation

Complementation in trans of defective IRES elements may be dueto various types of interactions. To get some insights into the influenceof RNA–RNA interactions on the complementation efficiency weprepared an IRES construct (termed E2) carrying nucleotide substitu-tions in the proximal part of the domain 3 that destabilized the base-paired structure of this region (Fig. 5A) and had a diminished RNA–RNA interaction capacity with domain 3 transcript (Fig. 5B). Thismutant, that is only 10-fold less active than the wild type IRES(Fig. 5C), exhibited a reduction in the capacity to complement thedefective IRES mutants, GUAG and CGCCC, in comparison to the wildtype IRES (a decrease from 11-fold to 2.6-fold in the IRES activityrescue was observed). Thus, defects in RNA–RNA interaction withindomain 3 can affect the efficiency of IRES trans-complementation.

Of interest, we noticed that the E2 mutant rescued the activity ofthe eIF4G-binding defective IRES (Fig. 5C). Relative to the efficiency ofprotein synthesis displayed by the mutant bicistronic constructcotransfected with the empty vector, a 12.8-fold increase wasobserved in the presence of the E2 IRES, whereas the wild type IRESinduced a 43-fold increase. These results suggested that a partialdefective IRES element with perturbation in the RNA structure of thecentral domain can complement, althoughmore weakly than the wildtype IRES, defects residing in domain 4 which are involved in theengagement of eIFs essential for IRES function.

Defects caused by impairment of IRES interactions with essentialtranslation components can be complemented in trans by transcriptswith defects in other IRES regions

To test the possibility that the complementing RNA is facilitating,or substituting, the defective one in performingmolecular interactionswith the components of the translation machinery, we tried to rescuethe activity of IRESmutants unable to interact with eIF4Gwith the alsohighly defective GUAG IRES, affected in the RNA structure of thecentral domain but retaining the capacity to interact with eIF4G(Fernandez-Miragall and Martinez-Salas, 2003). The rescue of IRESactivity was weaker than that observed with the wild type IRES, butclearly above the background observed with the empty vector as wellas that seenwith the HCV IRES (Fig. 3B), and reached values similar tothose rescued by the E2 IRES variant. An increase in IRES activity ofabout 10-fold was observed relative to the value obtained for the IRES

Fig. 5. (A) RNA structure of the E2 IRES mutant determined by RNA probing. Residues exposed to attack by dimethyl sulfate (DMS) or to ribonuclease T1 within the modified regionare shown. (B) RNA–RNA interaction of domain 3 with IRES regions. The indicated unlabeled transcript was incubated with either D3-wt or D3-E2 probe (12.5 nM) in binding buffercontaining 1mMMgCl2. Complexes were fractionated in 6% native gels in TBM buffer. For quantitative analysis of D3 self-interaction, the amount of retarded probewas normalized tothe intensity of the entire lane.White bars correspond to the intensity of the free probe, and gray bars to the intensity of the retarded probe using two concentrations of the unlabeleddomain 3 (light gray, 200 nM, dark gray 400 nM). (C) Rescue of IRES activity by transcripts encompassing the FMDV IRES with a reorganized stem 3 structure. The mutant bicistronicplasmid (25 ng/2×104 cells) carrying substitutions in the GNRAmotif (GUAG), or the RAAAmotif (CGCCC), or in domain 4 (4G-binding) was cotransfected with a plasmid (12.5 ng/2×104 cells) expressing the entire wild type IRES element (wt, black bars) or the mutant with a reorganized structure in the stem 3 region (E2, striped bars). IRES activity rescuewascalculated as in Fig. 2B.

226 P. Serrano et al. / Virology 388 (2009) 221–229

construct transfected with the empty vector (Fig. 6). Therefore,defects caused by the impairment of IRES interactions with essentialtranslation factors can be complemented in trans by IRES transcriptswith defects in other regions. This result demonstrates a division offunctions between the FMDV IRES domains.

Discussion

The RNA structure of viral IRES elements plays a crucial role ininternal initiation of translation (Lukavsky, 2009; Martinez-Salas,2008). From the structural point of view, it is well established that

distant interactions mediate RNA tertiary contacts strictly required fordicistrovirus IRES activity (Jan, 2006; Kieft, 2009; Nakashima andUchiumi, 2009). In picornavirus IRES elements it has been long knownthat structural domains are engaged in RNA–protein interactionsnecessary for IRES activity (Jang and Wimmer, 1990; Kolupaeva et al.,1998; Lopez de Quinto andMartinez-Salas, 2000; Luz and Beck, 1991).However, at least in the FMDV IRES, interaction with eIFs is necessarybut not sufficient to achieve IRES activity (Fernandez-Miragall et al.,2009). Accordingly, it has been suggested that picornavirus IRESelements have a modular organization with a specific distribution offunctions. In this hypothetical model the different domains of the IRES

Fig. 6. Rescue of the IRES mutant impaired in eIF4G binding by other defective elementswith altered RNA structure within domain 3. The mutant bicistronic plasmid (25 ng/2×104 cells) carrying substitutions AA312–313 to GC in domain 4 that abolishedinteraction with eIF4G was cotransfected with a plasmid (12.5 ng/2×104 cells)expressing the wild type IRES element (black bars) or mutants harboring either asubstitution in the GNRA motif (GUAG, striped bars), or the stem 3 reorganizedstructure (E2, gray bars). IRES activity rescue was calculated as in Fig. 2B.

227P. Serrano et al. / Virology 388 (2009) 221–229

perform distinct functions, all of them integrated within the process ofinternal initiation of protein synthesis. If this model proves to be true,it remains possible that some of these functions could be performed intrans by IRES transcripts.

To date, the precise role played by domain 3 during internalinitiation is still poorly understood. Here we have used a comple-mentation assay to further assess the relevance of the central domainin FMDV IRES activity. This domain has been described to self-interactvia RNA–RNA interactions in a very efficient manner, and althoughmore weakly, to form long-range RNA–RNA complex with the distaldomains 1–2 and 4–5 (Ramos andMartinez-Salas,1999). In addition, atertiary structural motif that resides in the apical region of thisdomain, and is an integral part of the IRES function, mediates distantRNA–RNA bridges within the central domain (Fernandez-Miragall andMartinez-Salas, 2003; Fernandez-Miragall et al., 2006; Serrano et al.,2007). The interactions observed in vitro between IRES domains, or itssubdomains, could provide the basis to explain the ability of theseRNAs to complement defective IRES elements in vivo, presumably bygenerating a physical link between the defective IRES and thetranslation machinery engaged by the active one.

We report here that the entire FMDV IRES transcript is able tocomplement severely defective molecules which display between 0.5and 1% of the wild type IRES activity, and are affected in the RNAorganization of the apical region of the central domain. Rescue of IRESactivity was partial, reaching about 10% of the wild type IRESefficiency. The relatively low levels of IRES rescue may be due to theabsence of selection during the rescue assay. No complementationwasobserved with the unrelated but also efficient HCV IRES, demonstrat-ing that IRES rescue was strictly dependent on the expression ofhomologous IRES transcript.

Although with lower efficiency than the entire IRES, the transcriptencompassing the wild type domain 3 alone can also complementdefects of the central domain. However, neither the proximal region ofthis domain (stem 3) nor domains 1–2 were able to rescue defectiveIRES mutants. Additionally, and in support of the implications of RNA-mediated contacts in the rescue of IRES activity, we have found that a

mutation affecting the capacity of the IRES to mediate RNA–RNAinteractions caused a decrease in the rescue of defective IRES.Therefore, the capacity of domain 3 to promote RNA–RNA interactionsin vitro is necessary but not sufficient to confer the complementationproperty to the FMDV IRES in transfected cells. Instead, a correct RNAstructure in this region in necessary to mediate complementation intrans.

The FMDV IRES element affected in its capacity to interact with theessential translation initiation factor eIF4Gwas also complemented bythe wild type IRES transcript. Remarkably, and although more weaklythan the wild type IRES, we show here that the eIF4G-bindingdefective IRES could be complemented by the also defective GUAGIRES transcript. This event could arise as the consequence of RNA-mediated contacts between defective molecules, each carrying adefect in a separate domain but having the complementing domainwith the right structural orientation and/or with the appropriateribonucleoprotein composition.

Our work confirms and extends previous studies that showed thattruncated regions of the EMCV IRES were able to complementdefective IRES elements as far as the complementing transcriptcontained at least the altered domain (Roberts and Belsham, 1997).Despite the similar property of EMCV and FMDV RNAs, IREScomplementation in trans is not a general property of IRES elements.For instance, mutant HCV IRES elements, ranging from 0.4 to 95.8% ofthe activity of the prototype HCV IRES, could not be complemented intrans by a wild-type HCV IRES element (Tang et al., 1999). Thisbehavior may reflect important differences in the mechanism drivinginternal initiation by this unrelated IRES element.

HCV and picornavirus IRES elements differ in primary sequence,secondary RNA structure and IRES-transacting factors requirement.Hepatitis C virus, an RNA virus belonging to the Flaviviridae family,initiates translation of its genomic RNA internally (Lukavsky, 2009).The HCV IRES spans 340 bases within the 5′UTR and its activitydepends on its structural organization (Honda et al., 1996), being theentire HCV IRES needed for activity (Otto and Puglisi, 2004). Assemblyof 48S complexes in vitro depends on the interaction with eIF3 andeIF2, but not eIF4G (Pestova et al., 1998). Thus, the HCV IRES issignificantly shorter than the picornavirus counterparts, it requiresless host cell factors for 48S complex assembly, and its RNA structurehas no apparent conservation with the FMDV IRES.

Taken together, our results show that defects caused by impair-ment of IRES interactions with essential translation components canbe complemented in trans by IRES transcripts with defects in otherregions, therefore, evidencing a division of functions between theFMDV IRES domains. The complementation ability of the FMDV IRESshown here may have important consequences for virus viability.

Materials and methods

Constructs

pGEM-based clones containing the full-length FMDV IRESsequence or the different domains have been previously described(Lopez de Quinto et al., 2001; Ramos and Martinez-Salas, 1999).Nucleotide substitutions in conserved motifs of the FMDV IRES weredescribed (Fernandez-Miragall and Martinez-Salas, 2003; Lopez deQuinto and Martinez-Salas, 1997). The HCV IRES construct waspreviously described (Lafuente et al., 2002).

The construct E2, bearing nucleotide substitutions in the stem ofdomain 3, was generated by PCR using oligonucleotides 5′-CAGTGCTGTTACTCGGGTCGCCATGG-3 (antisense) and 5′-CGAT-GAGTGGCAGGGCGGGGGC-3′ (sense) using as template the plasmidpBIC (Martinez-Salas et al., 1993). This PCR product was used in asecond PCR reaction with the primer 5′GGCCTTTCTTTATGTTTTT-GGCG-3′ to obtain the full-length FMDV IRES sequence, as described(Martinez-Salas et al., 1996).

228 P. Serrano et al. / Virology 388 (2009) 221–229

IRES activity rescue assays

Confluent BHK-21 cell monolayers grown in 24-well dishes weretransfected with the indicated plasmids, using the T7 RNA polymeraseexpression system as described (Martinez-Salas et al., 1993). For IRESrescue assays, BHK-21 cells were transfected with two plasmids, oneexpressing the defective IRES within a bicistronic RNA of the formCAT-IRES-luciferase and the second one, expressing solely thecomplementing IRES transcript (Fig. 2A). IRES activity was quantifiedas the expression of luciferase (IRES-dependent) normalized to that ofchloramphenicol acetyl transferase (CAT) (cap-dependent) and then,made relative to the value obtained for the wild type bicistronicconstruct, supplemented with the empty vector. The amount ofplasmid DNA expressing the complementing RNA (encompassingexclusively the IRES region) was varied between 3.25 and 50 ng/2×104 cells, using 25 ng/2×104 cells of the plasmid expressing thedefective bicistronic RNA. Appropriate amounts of the empty vectorpGEM3 were used to supplement the transfection mixture to provideequal amounts of plasmid DNA. Cell extracts were prepared 20 h posttransfection, and the activity of CAT and luciferase was determined(Lopez de Quinto et al., 2002). The ratio of Luc/CAT obtained in eachextract was normalized to the value obtained in the transfection withthe plasmid expressing the wild type IRES, pBIC, set to 100%. Therescued IRES activity reached a plateau above 12.5 ng/2×104 cells ofthe complementing IRES construct, being similar between 20 and100 ng. Thus, assays were carried out with 12.5 ng/2×104 cells, unlessotherwise stated.

In vitro transcription, gel shift, and RNA probing assays

Prior to RNA synthesis, plasmids were linearized using XhoI togenerate the IRES transcript, and SmaI for D3 and stem 3 transcripts.D1-2 and D4-5 transcripts were obtained from HindIII or XhoIlinearized templates. Transcription was performed for 1 h at 37 °Cusing 50 U of T7 RNA polymerase in the presence of 0.5 to 1 μg oflinearized DNA template, 50 mM DTT, 0.5 mM rNTPs and 20 U ofRNasin. When needed, RNA transcripts were uniformly labeled using[α32P]CTP (400 Ci/mmol) as described (Serrano et al., 2007).

For RNA–RNA interactions, uniformly 32P-labeled transcripts(12.5 nM) were incubated independently at 95 °C for 3 min,transferred to ice, and then mixed with the indicated unlabeled RNA(200–400 nM) denatured as above, in 50 mM sodium cacodylate pH7.5, 300 mM KCl, 1 mM MgCl2 (Serrano et al., 2006). The antisenseversion of the domains 1–2 (1–2 as), was used as a negative control.RNA–RNA complexes were allowed to form for 30 min at 37 °C andimmediately analyzed by electrophoresis in native polyacrylamidegels supplemented with 2.5 mM MgCl2. Electrophoresis wasperformed at 4 °C for 33 min at 180 V in TBM buffer (45 mM Tris,pH 8.3, 43 mM boric acid, 0.1 mMMgCl2) (Ramos and Martinez-Salas,1999). Dried gels were exposed for autoradiography and to aphosphorimager plate to quantify the intensity of the retarded bands.

RNA structure probing was performed using dimethyl sulfate(DMS) and ribonuclease T1 as described in (Brunel and Romby,2000; Fernandez-Miragall and Martinez-Salas, 2007). For primerextension, RNA was denatured by heating for 3 min at 95 °C.Annealing and extension of the labeled antisense primer was carriedout in 15 μl of reverse transcriptase (RT) buffer (20 mM Tris–HCl (pH7.5), 15 mM MgCl2, 100 mM NaCl, 0.1 mM EDTA, 1 mM DTT, 0.01%(v/v) NP-40, 50% (v/v) glycerol in the presence of 100 U ofSuperScript II RT and 1 mM each dNTPs for 1 h at 45 °C. Newlysynthesized cDNA products were analyzed in denaturing 6%acrylamide, 7 M urea gels (Fernandez-Miragall et al., 2006). Asequence ladder, prepared with the 5′-labeled antisense oligonucleo-tide used for primer extension, was loaded in parallel to identify theRT-extension products. The modified nucleotides were identified asthe base preceding one position the RT-stop.

Acknowledgments

We are grateful to GJ Belsham and C Gutierrez for helpfulsuggestions on the manuscript. We also thank Susana Barroso fortechnical assistance. This work was supported by grants BFU2005-00948, BFU2008-02159 and by an Institutional grant from FundaciónRamón Areces.

References

Andreev, D.E., Fernandez-Miragall, O., Ramajo, J., Dmitriev, S.E., Terenin, I.M., Martinez-Salas, E., Shatsky, I.N., 2007. Differential factor requirement to assemble translationinitiation complexes at the alternative start codons of foot-and-mouth disease virusRNA. RNA 13, 1366–1374.

Belsham, G.J., 2009. Divergent picornavirus IRES elements. Virus Res. 139, 184–193.Belsham, G.J., Martinez-Salas, E., 2004. Genome organisation, translation and replica-

tion of foot-and-mouth disease virus RNA. In: Domingo, E., Sobrino, F. (Eds.), Foot-and-Mouth Disease: Current Perspectives. Horizon Scientific Press, pp. 19–52.

Brunel, C., Romby, P., 2000. Probing RNA structure and RNA-ligand complexes withchemical probes. Methods Enzymol. 318, 3–21.

Drew, J., Belsham, G.J., 1994. trans complementation by RNA of defective foot-and-mouth disease virus internal ribosome entry site elements. J. Virol. 68, 697–703.

Fernandez-Miragall, O., Martinez-Salas, E., 2003. Structural organization of a viral IRESdepends on the integrity of the GNRA motif. RNA 9, 1333–1344.

Fernandez-Miragall, O., Martinez-Salas, E., 2007. In vivo footprint of a picornavirusinternal ribosome entry site reveals differences in accessibility to specific RNAstructural elements. J. Gen. Virol. 88, 3053–3062.

Fernandez-Miragall, O., Ramos, R., Ramajo, J., Martinez-Salas, E., 2006. Evidence ofreciprocal tertiary interactions between conserved motifs involved in organizingRNA structure essential for internal initiation of translation. RNA 12, 223–234.

Fernandez-Miragall, O., Lopez de Quinto, S., Martinez-Salas, E., 2009. Relevance of RNAstructure for the activity of picornavirus IRES elements. Virus Res. 139, 173–183.

Gutierrez, A., Martinez-Salas, E., Pintado, B., Sobrino, F., 1994. Specific inhibition ofaphthovirus infection by RNAs transcribed from both the 5′ and the 3′ noncodingregions. J. Virol. 68, 7426–7432.

Honda, M., Ping, L.H., Rijnbrand, R.C., Amphlett, E., Clarke, B., Rowlands, D., Lemon, S.M.,1996. Structural requirements for initiation of translation by internal ribosomeentry within genome-length hepatitis C virus RNA. Virology 222, 31–42.

Jan, E., 2006. Divergent IRES elements in invertebrates. Virus Res. 119, 16–28.Jang, S.K., Wimmer, E., 1990. Cap-independent translation of encephalomyocarditis

virus RNA: structural elements of the internal ribosomal entry site and involvementof a cellular 57-kD RNA-binding protein. Genes Dev. 4, 1560–1572.

Kieft, J.S., 2009. Comparing the three-dimensional structures of Dicistroviridae IGR IRESRNAs with other viral RNA structures. Virus Res. 139, 148–156.

Kolupaeva, V.G., Pestova, T.V., Hellen, C.U., Shatsky, I.N., 1998. Translation eukaryoticinitiation factor 4G recognizes a specific structural element within the internalribosome entry site of encephalomyocarditis virus RNA. J. Biol. Chem. 273,18599–18604.

Lafuente, E., Ramos, R., Martinez-Salas, E., 2002. Long-range RNA–RNA interactionsbetween distant regions of the hepatitis C virus internal ribosome entry siteelement. J. Gen. Virol. 83, 1113–1121.

Lopez de Quinto, S., Martinez-Salas, E., 1997. Conserved structural motifs located indistal loops of aphthovirus internal ribosome entry site domain 3 are required forinternal initiation of translation. J. Virol. 71, 4171–4175.

Lopez de Quinto, S., Martinez-Salas, E., 2000. Interaction of the eIF4G initiation factorwith the aphthovirus IRES is essential for internal translation initiation in vivo. RNA6, 1380–1392.

Lopez de Quinto, S., Lafuente, E., Martinez-Salas, E., 2001. IRES interaction withtranslation initiation factors: functional characterization of novel RNA contactswith eIF3, eIF4B, and eIF4GII. RNA 7, 1213–1226.

Lopez de Quinto, S., Saiz, M., de la Morena, D., Sobrino, F., Martinez-Salas, E., 2002. IRES-driven translation is stimulated separately by the FMDV 3′-NCR and poly(A)sequences. Nucleic Acids Res. 30, 4398–4405.

Lukavsky, P.J., 2009. Structure and function of HCV IRES domains. Virus Res. 139,167–172.

Luz, N., Beck, E., 1991. Interaction of a cellular 57-kilodalton protein with the internaltranslation initiation site of foot-and-mouth disease virus. J. Virol. 65, 6486–6494.

Martinez-Salas, E., 2008. The impact of RNA structure on picornavirus IRES activity.Trends Microbiol. 16, 230–237.

Martinez-Salas, E., Saiz, J.C., Davila, M., Belsham, G.J., Domingo, E., 1993. A singlenucleotide substitution in the internal ribosome entry site of foot-and-mouthdisease virus leads to enhanced cap-independent translation in vivo. J. Virol. 67,3748–3755.

Martinez-Salas, E., Regalado, M.P., Domingo, E., 1996. Identification of an essentialregion for internal initiation of translation in the aphthovirus internal ribosomeentry site and implications for viral evolution. J. Virol. 70, 992–998.

Martinez-Salas, E., Pacheco, A., Serrano, P., Fernandez, N., 2008. New insights intointernal ribosome entry site elements relevant for viral gene expression. J. Gen.Virol. 89, 611–626.

Nakashima, N., Uchiumi, T., 2009. Functional analysis of structural motifs indicistroviruses. Virus Res. 139, 137–147.

Otto, G.A., Puglisi, J.D., 2004. The pathway of HCV IRES-mediated translation initiation.Cell 119, 369–380.

229P. Serrano et al. / Virology 388 (2009) 221–229

Pacheco, A., Reigadas, S., Martinez-Salas, E., 2008. Riboproteomic analysis of polypep-tides interacting with the internal ribosome-entry site element of foot-and-mouthdisease viral RNA. Proteomics 8, 4782–4790.

Pacheco, A., Lopez de Quinto, S., Ramajo, J., Fernandez, N., Martinez-Salas, E., 2009. Anovel role for Gemin5 in mRNA translation. Nucleic Acids Res. 37, 582–590.

Pestova, T.V., Shatsky, I.N., Fletcher, S.P., Jackson, R.J., Hellen, C.U., 1998. A prokaryotic-like mode of cytoplasmic eukaryotic ribosome binding to the initiation codonduring internal translation initiation of hepatitis C and classical swine fever virusRNAs. Genes Dev. 12, 67–83.

Pilipenko, E.V., Pestova, T.V., Kolupaeva, V.G., Khitrina, E.V., Poperechnaya, A.N., Agol, V.I., Hellen, C.U., 2000. A cell cycle-dependent protein serves as a template-specifictranslation initiation factor. Genes Dev. 14, 2028–2045.

Ramos, R., Martinez-Salas, E., 1999. Long-range RNA interactions between structuraldomains of the aphthovirus internal ribosome entry site (IRES). RNA 5, 1374–1383.

Roberts, L.O., Belsham, G.J., 1997. Complementation of defective picornavirus internalribosome entry site (IRES) elements by the coexpression of fragments of the IRES.Virology 227, 53–62.

Saiz, J.C., Lopez de Quinto, S., Ibarrola, N., Lopez-Labrador, F.X., Sanchez-Tapias, J.M.,Rodes, J., Martinez-Salas, E., 1999. Internal initiation of translation efficiency indifferent hepatitis C genotypes isolated from interferon treated patients. Arch.Virol. 144, 215–229.

Saiz, M., Gomez, S., Martinez-Salas, E., Sobrino, F., 2001. Deletion or substitution of theaphthovirus 3′ NCR abrogates infectivity and virus replication. J. Gen. Virol. 82,93–101.

Semler, B.L., Waterman, M.L., 2008. IRES-mediated pathways to polysomes: nuclearversus cytoplasmic routes. Trends Microbiol. 16, 1–5.

Serrano, P., Rodriguez Pulido, M., Saiz, M., Martinez-Salas, E., 2006. The 3′ end of thefoot-and-mouth disease virus genome establishes two distinct long-range RNA–RNA interactions with the 5′ end region. J. Gen. Virol. 87, 3013–3022.

Serrano, P., Gomez, J., Martinez-Salas, E., 2007. Characterization of a cyanobacterialRNase P ribozyme recognition motif in the IRES of foot-and-mouth disease virusreveals a unique structural element. RNA 13, 849–859.

Sobrino, F., Saiz, M., Jimenez-Clavero, M.A., Nunez, J.I., Rosas, M.F., Baranowski, E., Ley, V.,2001. Foot-and-mouth disease virus: a long known virus, but a current threat. Vet.Res. 32, 1–30.

Stassinopoulos, I.A., Belsham, G.J., 2001. A novel protein-RNA binding assay: functionalinteractions of the foot-and-mouth disease virus internal ribosome entry site withcellular proteins. RNA 7, 114–122.

Stone, D.M., Almond, J.W., Brangwyn, J.K., Belsham, G.J., 1993. trans complementation ofcap-independent translation directed by poliovirus 5′ noncoding region deletionmutants: evidence for RNA–RNA interactions. J. Virol. 67, 6215–6223.

Tang, S., Collier, A.J., Elliott, R.M., 1999. Alterations to both the primary and predictedsecondary structure of stem-loop IIIc of the hepatitis C virus 1b 5′ untranslatedregion (5′UTR) lead to mutants severely defective in translation which cannot becomplemented in trans by the wild-type 5′UTR sequence. J. Virol. 73,2359–2364.

Vagnozzi, A., Stein, D.A., Iversen, P.L., Rieder, E., 2007. Inhibition of foot-and-mouthdisease virus infections in cell cultures with antisense morpholino oligomers. J.Virol. 81, 11669–11680.

Van Der Velden, A., Kaminski, A., Jackson, R.J., Belsham, G.J., 1995. Defective pointmutants of the encephalomyocarditis virus internal ribosome entry site can becomplemented in trans. Virology 214, 82–90.