A Stretch of 11 Amino Acids in the B-C Loop of the Coat Protein of Grapevine Fanleaf Virus Is

J OURNAL OF V IROLOGY,Aug.2010,p.7924–7933Vol.84,No.16 0022-538X/10/$12.00doi:10.1128/JVI.00757-10Copyright©2010,American Society for Microbiology.All Rights Reserved.A Stretch of11Amino Acids in the␤B-␤C Loop of the Coat Proteinof Grapevine Fanleaf Virus Is Essential for Transmission by theNematode Xiphinema indexᰔ†Pascale Schellenberger,1,2Peggy Andret-Link,1Corinne Schmitt-Keichinger,2Marc Bergdoll,2 Aure´lie Marmonier,1Emmanuelle Vigne,1Olivier Lemaire,1Marc Fuchs,3Ge´rard Demangeat,1and Christophe Ritzenthaler2*Institut National de la Recherche Agronomique,UMR1131INRA/Universite´de Strasbourg,28Rue de Herrlisheim,68021Colmar, France1;Institut de Biologie Mole´culaire des Plantes du CNRS,Universite´de Strasbourg,12Rue du Ge´ne´ral Zimmer,67084Strasbourg, France2;and Department of Plant Pathology,Cornell University,New York State Agricultural Experiment Station,Geneva,New York144563Received9April2010/Accepted21May2010Grapevine fanleaf virus(GFLV)and Arabis mosaic virus(ArMV)from the genus Nepovirus,family Secoviridae,cause a severe degeneration of grapevines.GFLV and ArMV have a bipartite RNA genome and are transmittedspecifically by the ectoparasitic nematodes Xiphinema index and Xiphinema diversicaudatum,respectively.Thetransmission specificity of both viruses maps to their respective RNA2-encoded coat protein(CP).To furtherdelineate the GFLV CP determinants of transmission specificity,three-dimensional(3D)homology structuremodels of virions and CP subunits were constructed based on the crystal structure of Tobacco ringspot virus,thetype member of the genus Nepovirus.The3D models were examined to predict amino acids that are exposed atthe external virion surface,highly conserved among GFLV isolates but divergent between GFLV and ArMV.Five short amino acid stretches that matched these topographical and sequence conservation criteria wereselected and substituted in single and multiple combinations by their ArMV counterparts in a GFLV RNA2cDNA clone.Among the21chimeric RNA2molecules engineered,transcripts of only three of them inducedsystemic plant infection in the presence of GFLV RNA1.Nematode transmission assays of the three viablerecombinant viruses showed that swapping a stretch of(i)11residues in the␤B-␤C loop near the icosahedral3-fold axis abolished transmission by X.index but was insufficient to restore transmission by X.diversicaudatumand(ii)7residues in the␤E-␣B loop did not interfere with transmission by the two Xiphinema species.Thisstudy provides new insights into GFLV CP determinants of nematode transmission.The transmission of a virus from one plant to another is a key step in the virus life cycle.For a majority of plant viruses, transmission is dependent upon vectors,mostly arthropods, soil-inhabiting nematodes,and plasmodiophorids(13,15).A successful transmission depends on the competency of the vec-tor to acquire virions,bind them to specific retention sites,and release them in a recipient plant.Virus transmission by a vector is often characterized by some degree of specificity;it can be broad or narrow,but it is a prominent feature for numerous plant viruses and their vectors(2).The viral deter-minants engaged in vector transmission involve the coat pro-tein(CP)that either binds directly to vector ligands or associ-ates with nonstructural viral proteins,like a helper component (HC),to create a molecular bridge between virions and the vector(25,44).For both strategies,only a very limited number of amino acids on either the CP or the HC contribute to the transmission process and to vector specificity.For the aphid-transmitted Cucumber mosaic virus(CMV;genus Cucumovirus,family Cucumoviridae),the CP is the sole viral determinant of transmission(10,12),and a negatively charged loop of eight residues located on the surface of virions is directly involved in vector interactions(19,26).For potyvi-ruses such as Tobacco vein mottling virus(TVMV),the CP and HC are both involved in transmission by aphids.The amino acid motifs DAG located near the N terminus of the CP,KITC located near the N terminus of the HC-Pro,and PTK located into the central part of the HC-Pro are essential for transmis-sion(35).For Cauliflower mosaic virus(CaMV;genus Cauli-movirus,family Caulimoviridae),a single amino acid change in protein P2,which acts as HC,is sufficient to abolish aphid transmission(24).In the case of nematode-transmitted viruses, a single mutation in the HC protein2b of Pea early-browning virus(PEBV;genus Tobravirus,unassigned family)is sufficient to prevent transmission(38).Similarly,a15-amino-acid dele-tion at the C terminus of the CP abolishes transmission of PEBV(22).For members of the genus Nepovirus in the family Secoviridae,little information is available on viral determi-nants that govern nematode transmission(21).Transmission assays of pseudorecombinants between Raspberry ringspot virus(RpRSV)and Tobacco black ring virus(TBRV)showed the involvement of RNA2-encoded proteins and suggested a role of the CP in transmission specificity(reviewed in ref-erence3).Based on sequence alignments,the residue in position219of the RpRSV CP was proposed to be exposed*Corresponding author.Mailing address:Institut de Biologie Mo-le´culaire des Plantes du CNRS,Universite´de Strasbourg,12Rue duGe´ne´ral Zimmer,67084Strasbourg,France.Phone:33388417257. Fax:33388614442.E-mail:ritzenth@unistra.fr.†Supplemental material for this article may be found at http://jvi/.ᰔPublished ahead of print on2June2010.7924 on December 9, 2015 by CHINA AGRICULTURE UNIVERSITY / Downloaded fromat the surface of particles and involved in nematode trans-mission(33).However,no experimental evidence is avail-able to confirm these hypotheses.Thus,our knowledge of the viral determinants of nepovirus transmission is rather scarce.Grapevine fanleaf virus(GFLV)and Arabis mosaic virus (ArMV),two closely related species of the genus Nepovirus, family Secoviridae(31,32),are responsible for a severe degen-eration of grapevines that occurs in most vineyards worldwide (3).GFLV and ArMV have a bipartite RNA genome and share a similar genome organization,as well as a relatively high level of sequence similarity(39).The RNA1of both viruses encodes replication and protein maturation functions,and the RNA2encodes a protein involved in RNA2replication,which is referred to as a homing protein(2A HP),the movement protein(2B MP)and the coat protein(2C CP)(3;see also Fig.4). The nematode transmission of GFLV and ArMV illustrates the type of extreme specificity that can exist between vectors and viruses(8).GFLV is transmitted exclusively by Xiphinema index and ArMV is transmitted specifically by X.diversicauda-tum.The transmission specificity of GFLV(4)and ArMV(23) maps to their respective2C CP,but no information is available on2C CP residues involved in vector interactions.The main objective of our study was to delineate2C CP amino acids involved in GFLV transmission by X.index and advance our knowledge of molecular features of nepovirus transmission.Three-dimensional(3D)homology-based mod-els of GFLV virions and CP subunits were constructed by using the3.5-Åcrystal structure of Tobacco ringspot virus(TRSV)as a template(9).TRSV is the type member of the genus Nepo-virus to which GFLV and ArMV belong(32)and is transmitted by the nematode X.americanum.From our structural models, residues predicted to be exposed at the external capsid surface, highly conserved among GFLV isolates but divergent between GFLV and ArMV isolates,were identified and selected for a reverse-genetics approach to determine their involvement in vector transmission.Our results indicate that a stretch of11res-idues predicted in the␤B-␤C loop of the CP B domain is essential for X.index-mediated transmission of GFLV.To our knowledge, this represents thefirst description of structural amino acids es-sential for nepovirus transmission by nematodes.MATERIALS AND METHODSVirus strains.GFLV strain F13(41)and ArMV strain S(16)were isolated from naturally infected grapevines and propagated in Chenopodium quinoa,a systemic host for both viruses.The sequence of the two genomic RNAs of both virus strains is determined(20,28,34;C.Ritzenthaler et al.,unpublished). Full-length cDNA clones of GFLV-F13RNA1(plasmid pMV13)and RNA2 (plasmid pVec2ABC)are available for in vitro synthesis of transcripts(40). ArMV-S was used as a control in nematode transmission assays,and a cDNA of ArMV-S RNA2-U(20)was used in mutagenesis experiments.GFLV3D structure model building.A protein/protein BLAST(1)search for potential structural templates was conducted using the query GFLV-F13CP sequence(GI25013721,504amino acids)against sequences available in the Protein Data Bank.CLUSTAL X(36)alignment was restricted to GFLV-F13, ArMV-S(GI757527,505amino acids)and TRSV(GI3913279;513amino acids) sequences.The structure described in TRSV1A6C was used as a template for GFLV model building using MODELLER(30).Alignment was then refined by hand to accommodate structural constraints revealed by the3D model.Putative surface residues of GFLV were considered to be those that aligned with the surface residues identified for TRSV by Chandrasekar and Johnson(9).Hex-amers were built rather than isolated monomers because virus capsids are highly constrained by high symmetry,extensive contacts between monomers,and the fact that approximately one-third of the total accessible surface residues are predictably buried by direct neighbors.The GFLV CP amino acid stretches presumably involved in transmission by X. index were identified by combining information from the GFLV capsid3D model,the surface residues identified by Chandrasekar and Johnson(9),and the alignment of219GFLV and20ArMV CP sequences(see Table S1in the supplemental material).Sequences were aligned with AlignX(Vector NTI). Development of chimeric GFLV RNA2.Plasmid pVec2ABC carrying a full-length cDNA copy of GFLV-F13RNA2was used as a template to develop chimeric2C CP genes(6).For cloning purposes,an Acc65I restriction site was introduced into pVec2ABC from positions2678to2683(nucleotide positions are given according to the GFLV-F13RNA2sequence,GenBank accession no. NC_003623)by site-directed PCR-mediated mutagenesis.The plasmid gener-ated was named pVec Acc65I2ABC.Introducing Acc65I modified the nucleotide but not the amino acid composition of the wild-type GFLV-F132C CP. Previous work showed that functional GFLV/ArMV RNA2recombinants can be produced,providing a functional interaction between proteins2B MP and2C CP is maintained for cell-to-cell movement(5,6).Mutations affecting regions R1(nucleotides[nt]2284to2302),R2(nt2609to2640),R4(nt2819 to2842),and R5(nt2937to2965)of the GFLV2C CP gene were produced by primer overlap extension mutagenesis(14),using the primers described in Table S1in the supplemental material.The mutation affecting region R3(nt2666to 2677)only required one PCR step because of the presence of the Acc65I restriction site into the reverse primer.Each PCR was carried out as described previously(4),except that the annealing time was reduced to45s at60°C and the elongation time was adapted to the size of the PCR products.DNA amplicons obtained by PCR were digested with appropriate restriction enzymes (see Table S1in the supplemental material)and cloned into the corresponding sites in pVec Acc65I2ABC to yield pVec Acc65I2ABC G1,pVec Acc65I2ABC G2, pVec Acc65I2ABC G3,pVec Acc65I2ABC G4,and pVec Acc65I2ABC G5(a“G”as a subscript indicates the GFLV origin of the2C CP,and1to5indicate the mutated regions R1to R5,respectively).For simplicity,transcripts and recombinant viruses derived from this series of constructs are referred to as G1,G2,G3,G4, and G5.To produce mutants carrying combinations of regions R1,R2,and R3,chi-meric plasmids pVec Acc65I2ABC G1and pVec Acc65I2ABC G2were used as the matrix for a second primer overlap—or single—PCR mutagenesis,and the resulting PCR fragments were cloned into pVec Acc65I2ABC,using NgoMIV and Acc65I to produce pVec Acc65I2ABC G12,pVec Acc65I2ABC G13and pVec Acc65I2ABC23(12,13,and23as subscripts indicate the combination of the mutated regions1and2,1and3,and2and3,respectively).For simplicity,transcripts and recombinant viruses derived from this second se-ries of constructs are referred to as G12,G13,and G23.To produce mutants carrying further combinations of swapped regions,mu-tant plasmids pVec Acc65I2ABC G1,pVec Acc65I2ABC G2,pVec Acc65I2ABC G3, pVec Acc65I2ABC G4,and pVec Acc65I2ABC G5were digested with appropriate re-striction enzymes(see Table S2in the supplemental material),and restriction fragments containing the mutated region(s)were cloned sequentially into plas-mids encoding a GFLV2C CP gene already carrying a mutation(see Table S2in the supplemental material).For simplicity,transcripts and recombinant viruses derived from this third series of constructs are referred to as G24,G25,G45,and G245.Plasmid pVec2AB U-4-9C was engineered earlier(4)from pVec2AB U9C(6)to produce chimeric RNA2constructs for which gene2C CP is of GFLV origin and gene 2B MP is of ArMV origin,except amino acids in positionsϪ2andϪ6upstream of the R/G cleavage site that remain of GFLV origin(4)in order to maintain a functional interaction between proteins2B MP and2C CP for systemic plant infection(6).Plas-mid pVec2AB U-4-9C was renamed in the present study as pVec2AB A C G(an“A”as a subscript indicates the chimeric nature of protein2B MP).The complete CP coding sequence of G1,G2,G4,G12,G24,G25,G45,and G245was cloned into plasmid pVec2AB A C G by using the appropriate restriction enzymes(see Table S2in the supplemental material)to produce pVec2AB A C G1,pVec2AB A C G2, pVec2AB A C G4,pVec2AB A CG12,pVec2AB A C G24,pVec2AB A C G25, pVec2AB A C G45,and pVec2AB A C G245,respectively.For simplicity,tran-scripts and recombinant viruses derived from this third series of constructs are referred to as AG1,AG2,AG4,AG12,AG24,AG25,AG45,and AG245. Plasmid pVec2AB U-6C U(4)was used in the present study to introduce CP gene region R2of GFLV origin in an ArMV CP gene and renamed as pVec2AB A C A(an“A”as a subscript indicates the chimeric nature of protein 2B MP and the ArMV nature of protein2C CP).Mutagenesis of CP region R2and cloning into pVec2AB A C A to yield pVec2AB A C A2was performed as described in Table S1in the supplemental material.For simplicity,transcripts derived from this construct are referred to as AA2.V OL.84,2010COAT PROTEIN DETERMINANTS OF GFLV TRANSMISSION7925on December 9, 2015 by CHINA AGRICULTURE UNIVERSITY /Downloaded fromThe integrity of the21GFLV chimeric RNA2clones produced for the present study was verified by DNA sequencing.In vitro transcription of GFLV cDNA clones and mechanical inoculation of plants.GFLV RNA1and RNA2,as well as GFLV chimeric RNA2,were ob-tained by in vitro transcription(6)and were mechanically inoculated into C. quinoa.Recombinant viruses were propagated in Nicotiana benthamiana by using crude sap of infected C.quinoa leaves as inoculum.Virus infection was assessed in uninoculated apical leaves of test plants two to3weeks postinoculation by double-antibody sandwich(DAS)-enzyme-linked immunosorbent assay(ELISA) with␥-globulins specific to GFLV and ArMV(42).Samples were considered positive if their optical densities at405nm(OD405s)were at least three times those of healthy controls after120min of substrate hydrolysis.Functionality of GFLV chimeric RNA2in Chenopodium quinoa protoplasts. The functionality of chimeric RNA2transcripts was evaluated by transfecting C. quinoa protoplasts in the presence of GFLV RNA1transcripts.Viral RNA replication and encapsidation were assessed by RNase protection experiments, as described previously(6),except that riboprobes used for Northern hybridiza-tion corresponded to nt4623to5810of RNA1(5Јhalf of gene1E POL)and nt233 to1006of RNA2(gene2A HP).Nematode transmission tests.Nematode transmission assays were performed with N.benthamiana for X.index(4)and C.quinoa for X.diversicaudatum(23). Transmission assays used a minimum offive to six plants and were performed at least twice,except for G1and AG1,which were tested only once.Synthetic GFLV,i.e.,RNA derived from pMV13and pVec Acc65I2ABC,referred to as 2ABC,and wild-type ArMV-S were used as controls in transmission assays with X.index and X.diversicaudatum.The presence of GFLV and ArMV was assessed in leaves of source plants by DAS-ELISA(4,23).Detection of GFLV and ArMV in nematodes.The presence of GFLV and ArMV was verified in nematodes by reverse transcription-PCR(RT-PCR)with a GFLV/ArMV consensus reverse primer EVAGR(5Ј-GGCAAGTGTGTCCA AAGGAC-3Ј)and a GFLV-specific forward primer GF(5Ј-ATGTGGAAGAG GACGGAAGT-3Ј)or an ArMV-specific forward primer AF(5Ј-GTTACATC GAGGAGGATG-3Ј)to amplify860-and849-bp specific fragments,respectively (11).Characterization of GFLV chimeric2C CP gene progeny.The progeny of chi-meric GFLV RNA2was characterized in infected plants by immunocapture (IC)/RT-PCR(5,6)with the primers208(see Table S1in the supplemental material)and398(5Ј-TGGGARARYRTNGAGGAAC-3Ј)that allow complete sequencing of the CP open reading frame.Sequences were analyzed with Con-tigExpress(Vector NTI).RESULTSConstruction of a GFLV homology-based3D model and search for surface residues potentially involved in X.index transmission.To identify GFLV2C CP amino acids involved in the transmission by X.index,we hypothesized that candidate residues are likely exposed at the external surface of virions, which are different between GFLV and ArMV isolates but highly conserved among GFLV isolates.The latter criterion was not considered absolute because no information is avail-able on nematode transmissibility of most of the219GFLV and20ArMV variants for which CP sequences are available in GenBank(see Table S1in the supplemental material).In the absence of a resolved structure for GFLV and ArMV, a3D structural model of GFLV virions was constructed by examining Protein Data Bank for potential structural tem-plates using the query GFLV-F13CP sequence.The closest structural template to the GFLV CP was the CP of TRSV (PDB-ID:1A6C)(9)with25%sequence identity,43%se-quence similarity,and a score of219positive residues out of 513for the crystal structure.CLUSTAL X alignments between GFLV-F13,ArMV-S,and TRSV were generated as indicated in Materials and Methods(Fig.1).This alignment slightly differed from the one described previously(9).We considered putative surface residues of GFLV to be those that aligned with the surface residues identified for TRSV by Chandrasekar and Johnson(9).3D homology models of GFLV virion and CP subunit constructed from the crystal structure of TRSV are presented in Fig.2.The virion icosahedral lattice is of a pseudo Tϭ3symmetry and one CP consists of three trapezoid shaped ␤-barrel domains referred to as the C,B,and A domains from the N to the C termini,as shown previously for TRSV(9).Of the156nonconserved residues between GFLV-F13and ArMV-S,59amino acids were predicted to be exposed at the external surface of the CP(highlighted in gray in Fig.1),and nearly half of those(27of the59)clustered within the central B domain.Considering this high number of divergent surface residues,we restricted our reverse-genetics approach to areas with the highest concentration of divergent residues rather than to single isolated residues.Five regions(R1to R5)rang-ing from4to11residues and displaying between0(region R3) and57%(region R4)amino acid sequence identity between GFLV and ArMV were identified and selected for mutagene-sis experiments(Fig.3).Thefive candidate regions all located to predicted loops connecting␤sheets within the B domain, except region R1that mapped to the C domain(Fig.1and2). Thefive CP regions were highly conserved within219GFLV or 20ArMV variants(Fig.3).When displayed in GFLV CP tertiary and quaternary structure models,thefive regions mapped close together,with R1closest to R3and R2closest to R5(Fig.2).Thefive regions surrounded a small depression between the5-fold and3-fold axes of the virion(Fig.2).Im-portantly,no direct contact of the selected regions between the different subunits could be deduced from our3D models, suggesting that the selected regions are unlikely to be in-volved in the CP-CP interactions necessary for virion assem-bly or stability.Engineering of recombinant viruses.A reverse-genetics ap-proach was used to investigate the involvement of GFLV can-didate CP regions R1to R5in nematode transmission.Chi-meric CP genes were obtained by substituting GFLV residues within regions R1to R5by their ArMV counterparts using site-directed mutagenesis of full-length infectious cDNA clones of GFLV RNA2(40).Regions were substituted indi-vidually,in pairs or in triple combination(Fig.4B).In addition, to circumvent possible defects in systemic movement of recom-binant viruses in planta(5,6),recombinants carrying single or combinations of ArMV CP regions R1to R5were also gener-ated in an infectious cDNA clone of GFLV RNA2for which the GFLV2B MP gene was replaced by its ArMV counterpart (Fig.4C),except for Leu343(positionϪ6upstream of the R/G cleavage site between proteins2B MP and2C CP)and Val347 (positionϪ2upstream of the R/G cleavage site)of protein 2B MP that remained of GFLV origin to maintain infectivity(4, 6).Altogether,21CP recombinants were generated,12in a full GFLV RNA2genetic background(Fig.4B),eight in a partial GFLV RNA2genetic background that included an ArMV protein2B MP(Fig.4C),and one in a partial GFLV RNA2 genetic background that included ArMV proteins2B MP and 2C CP(Fig.4D).Infectivity assays of recombinant viruses in planta.The in-fectivity of in vitro-synthesized GFLV RNA1and chimeric RNA2transcripts and the ability of the corresponding recom-binant viruses to cause systemic plant infection is essential to obtaining infected plant roots for nematode transmission as-says.Of the21GFLV recombinants tested,only recombinants7926SCHELLENBERGER ET AL.J.V IROL.on December 9, 2015 by CHINA AGRICULTURE UNIVERSITY /Downloaded fromG1,AG1,G2,and AG2caused a systemic infection in planta with typical GFLV symptoms and DAS-ELISA-positive unin-oculated apical leaves (Table 1).Recombinants exhibited a 2-to 5-day delay in symptom development compared to synthetic wild-type GFLV,i.e.,2ABC.However,this difference nearly vanished upon mechanical reinoculation of C.quinoa with crude sap from initially infected plant tissue,except for recom-binant G2,which,for unknown reasons,remained recalcitrant to reinoculation in C.quinoa and N.benthamiana and was therefore omitted from further studies.Recombinant viruses G1,AG1,and AG2were transferred successfully to N.benthamiana by mechanical inoculation,and the genetic sta-FIG.1.Map of GFLV predicted surface residues potentially involved in X.index transmission.Refined CLUSTAL X alignment of the CP residues of TRSV (GI 3913279),GFLV-F13(GI 25013721),and ArMV-S (GI 757527).The secondary structure of the TRSV atomic model is shown below the alignment.Surface residues of TRSV identified previously (9)are indicated by horizontal bars above the sequences.Predicted surface residues that are different between GFLV and ArMV are highlighted in gray.Boxes delineate regions 1(R1),2(R2),3(R3),4(R4),and 5(R5)used for inverse genetics studies because corresponding predicted surface residues are poorly conserved between GFLV and ArMV.V OL .84,2010COAT PROTEIN DETERMINANTS OF GFLV TRANSMISSION 7927on December 9, 2015 by CHINA AGRICULTURE UNIVERSITY/Downloaded frombility and integrity of their RNA2progeny was confirmed by IC/RT-PCR and sequencing of the CP gene.Importantly,nei-ther reversion to wild-type GFLV sequences nor compensatory mutations in the 2C CP were ever observed.Effect of CP mutations on encapsidation.The defect in in planta infectivity observed for most recombinant viruses engi-neered in the present study could result from deficiencies in replication,encapsidation,and/or cell-to-cell movement.To discriminate between these hypotheses,C.quinoa protoplasts were transfected with viral transcripts,and total RNA ex-tracted 72h posttransfection was analyzed by Northern hybrid-ization using GFLV RNA1and RNA2riboprobes.The results showed that all chimeric RNAs were competent for replication (Table 1).After replication in protoplasts,the ability of chi-meric protein 2C CP to encapsidate viral RNA was assessed under conditions that allows to distinguish RNase-protected,encapsidated viral RNA (P condition,PIPES buffer)from to-tal,encapsidated,and nonencapsidated viral RNA (T condi-tion,TLES buffer)as described previously (6).As expected,under T condition,progeny viral RNA was detected for all GFLV RNA1and RNA2combinations (Fig.5and Table 1),indicating that loss of infectivity was not due to replication.Under P condition,the only single recombinants in which progeny viral RNA could be detected were those affecting regions R1,R2,and R4,independently of the origin of the 2B MP gene (G1,G2,G4,AG2,and AG4,Fig.5and Table 1),indicating encapsidation of the corresponding viral genomic RNAs.In contrast,no viral RNA protection was observed for recombinants G3and G5as for mutant 2AB for which the CP gene is deleted (Fig.5).RNase protection assays further indi-cated a deficiency in viral RNA encapsidation for all double or triple recombinants,including the ones combining regions that individually could be exchanged with no effect on encapsida-tion,i.e.,G12,G24,and AG24(Fig.5and Table 1).Finally,no protected RNA could be detected for recombinant AA2(Fig.5),showing that even if the ArMV region R2could be placed in a GFLV 2C CP protein without effect on encapsi-dation (recombinant G2and AG2),the reverse swapping was deleterious.Effect of CP mutations on transmission by nematodes.The transmissibility of recombinant viruses that induced systemic plant infection,i.e.,G1,AG1,and AG2,by X.index and X.diversicaudatum was evaluated (4,5).Recombinant AG2was not transmitted by any of the two Xiphinema species,whereasFIG.2.Homology-based model of GFLV and 2C CP regions targeted for mutagenesis.(A)Ribbon representation of the GFLV CP deduced from the TRSV crystal structure (PDB ID 1A6C).The predicted A,B,and C structural domains are indicated in blue,red,and green,respectively.The five regions selected for mutational analysis due to their predicted surface location and divergence between GFLV and ArMV appear in yellow.Regions R2,R3,R4,and R5are located in loops in the B domain,whereas region R1is located in the C domain ␤E-␤F loop.(B)A space-filling model of GFLV particle quaternary structure viewed down the icosahedral 2-fold axis (top)and GFLV 2C CP subunit tertiary structure (bottom)showing the close proximity of regions R1to R5.The icosahedral 5-fold,3-fold,and 2-fold axes (labeled as 5,3,and 2,respectively)are labeled for one icosahedral asymmetric unit (top).The figures were generated using PYMOL ().7928SCHELLENBERGER ET AL.J.V IROL .on December 9, 2015 by CHINA AGRICULTURE UNIVERSITY/Downloaded fromrecombinants G1and AG1,like synthetic GFLV,were trans-mitted by X.index but not by X.diversicaudatum (Table 2).ArMV-S was the only virus transmitted by X.diversicaudatum (Table 2).Transmission rates varied between 67and 100%,which is in complete agreement with previous reports on GFLV and ArMV transmission rates (4,5).Such differences were statistically not sufficient to indicate possible variations in transmission efficiency between transmitted isolates.Analysis of the RNA2progeny of recombinants G1and AG1in infected bait plants by IC/RT-PCR and sequencing indicated no change in nucleotide sequence relative to the corresponding cDNA clones,confirming the genetic stability of the recombinants.These results are consistent with the notion that CP region R2is involved in GFLV transmission by X.index and CP region R1is not essential for transmission specificity.Effect of CP mutations on virus acquisition and retention by nematodes.The deficiency in nematode-mediated virus trans-mission could result from a lack of virus acquisition during the acquisition access period (AAP)or a failure of vectors to bind and release virus particles during the inoculation access period (IAP).To verify that nematodes fed on infected roots and ingested viruses,nematode specimens were randomly collected after the AAP and tested for virus presence by RT-PCR.Both vectors,X.index and X.diversicaudatum ,ingested recombi-nants G1,AG1,AG2,synthetic GFLV,and wild-type ArMV independently of their transmission competency (Fig.6).These results rule out the possibility that transmission failed due to a lack of virus acquisition by nematodes.Similarly,nematodes were tested for virus presence after the IAP to examine retention competency.Recombinants G1and AG1were detected in X.index ,while recombinant AG2was unde-tectable in the two nematode species (Fig.6).In agreement with the specific transmission of GFLV and ArMV (5),syn-thetic GFLV was only detected in X.index and wild-type ArMV in X.diversicaudatum (Fig.6).These RT-PCR results were consistent with our transmission data (Table 2)and con-firmed that CP region R2is required for GFLV retention in X.index ,whereas CP region R1does not seem essential for the specific virus retention by nematodes.In summary,CP region R2is likely exposed at the external surface of virions where it functions as a determinant for GFLV transmission by X.index .Unfortunately,we were not able to test CP region R2in the transmission of ArMV by X.diversicaudatum ,due to defect in encapsidation of the recip-rocal recombinant AA2(Table 1,Fig.5).The smaller CP gene region R1,which is probably also located at the external sur-face of particles,had no effect either on encapsidation or transmission of the virus.DISCUSSIONLimited information is available on viral determinants re-sponsible for nematode-mediated transmission of nepoviruses.Previous studies showed that the specific transmission of GFLV and ArMV by X.index and X.diversicaudatum ,respec-tively,is determined by their respective CPs (4,23).The iden-tification of CP residues involved in transmission has been hampered by unsuccessful attempts at isolating natural GFLV or ArMV isolates that are deficient in nematode transmission and by a lack of resolved virion structure.In the present study,FIG.3.Logographical representations of the amino acid sequence conservation within CP regions R1to R5of GFLV and ArMV.The sequences of 219GFLV and 20ArMV variants (see Table S1in the supplemental material)were used to create a logographical representation.The different residues at each position in the GFLV CP amino acid sequences (top of each panel)and the ArMV CP amino acid sequences (bottom of each panel)are scaled according to their frequency.The CP amino acid positions are given on the x axis of each panel.Black stars indicate residues that differ between GFLV and ArMV.The black line at the end of GFLV region R5(bottom right panel)represents a gap in the GFLV and ArMV 2C CP sequence alignment.Residues are colored according to their biochemical properties:positively charged in blue;negatively charged in red;hydrophobic in black;and polar uncharged in green and the proline,glycine,and cysteine in orange.CP sequence alignments were created by using AlignX (Vector NTI),and logo representations were obtained by using Weblogo (/logo.cgi).V OL .84,2010COAT PROTEIN DETERMINANTS OF GFLV TRANSMISSION 7929on December 9, 2015 by CHINA AGRICULTURE UNIVERSITY/Downloaded from。