Interaction of a Plant Virus-Encoded Protein with the Major Nucleolar Protein Fibrillarin Is Required for Systemic Virus Infection

Interaction of a plant virus-encoded protein with the major nucleolar protein fibrillarin is required for systemic virus infection

Sang Hyon Kim†, Stuart MacFarlane†, Natalia O. Kalinina†‡, Daria V. Rakitina†‡, Eugene V. Ryabov§, Trudi Gillespie†, Sophie Haupt†, John W. S. Brown†, and Michael Taliansky†¶

†Scottish Crop Research Institute, Invergowrie, Dundee DD2 5DA, United Kingdom; ‡A. N. Belozersky Institute of Physico-Chemical Biology, Moscow State University, Moscow 119992, Russia; and §Horticulture Research International, University of Warwick, Wellesbourne, Warwick CV35 9EF, United Kingdom

Communicated by Bryan D. Harrison, Scottish Crop Research Institute, Dundee, United Kingdom, May 17, 2007 (received for review April 4, 2007) The nucleolus and specific nucleolar proteins are involved in the life Umbraviruses have RNA genomes and differ from most other cycles of some plant and animal viruses, but the functions of these viruses in that they do not encode a coat protein (CP) and so do not proteins and of nucleolar trafficking in virus infections are largely produce conventional virus particles in infected plants (15, 16). unknown. The ORF3 protein of the plant virus, groundnut rosette Nevertheless, they accumulate and spread efficiently within the virus (an umbravirus), has been shown to cycle through the infected plant; their lack of a CP is compensated for by the ORF3 nucleus, passing through Cajal bodies to the nucleolus and then protein. This protein fulfils umbraviral functions that are normally exiting back into the cytoplasm. This journey is absolutely required provided by the CPs of other plant viruses, such as long-distance for the formation of viral ribonucleoprotein particles (RNPs) that, movement of viral RNA through the phloem (17, 18). themselves, are essential for the spread of the virus to noninocu- The GRV ORF3 protein interacts with viral RNA in vivo to form lated leaves of the shoot tip. Here, we show that these processes filamentous RNP particles, which have elements of a regular helical rely on the interaction of the ORF3 protein with fibrillarin, a major structure but not the uniformity typical of virus particles (16). The nucleolar protein. Silencing of the fibrillarin gene prevents long- RNPs accumulate in cytoplasmic inclusions that are the form in distance movement of groundnut rosette virus but does not affect which the virus is thought to move through the phloem to cause viral replication or cell-to-cell movement. Repressing fibrillarin systemic infection (16). In addition to its presence in the cytoplasm, production also localizes the ORF3 protein to multiple Cajal body- the ORF3 protein is able to traffic into the nucleus, predominantly like aggregates that fail to fuse with the nucleolus. Umbraviral targeting the nucleolus (19, 20). The presence of the ORF3 protein ORF3 protein and fibrillarin interact in vitro and, when mixed with in the nucleolus was unexpected, because the entire infection cycle umbravirus RNA, form an RNP complex. This complex has a fila- of GRV and other umbraviruses was previously considered to be mentous structure with some regular helical features, resembling restricted to the cytoplasm. ORF3 proteins contain two conserved the RNP complex formed in vivo during umbravirus infection. The domains: an arginine-rich sequence (positions 108–122; R-rich filaments formed in vitro are infectious when inoculated to plants, domain) and a leucine-rich region (amino acids 148–156; L-rich PLANT BIOLOGY and their infectivity is resistant to RNase. These results demon- domain) (Fig. 1) (16, 20). The R-rich domain is involved in nuclear strate previously undescribed functions for fibrillarin as an essen- import, and the L149 residue, in addition to the R-rich domain, is tial component of translocatable viral RNPs and may have impli- essential for nucleolar targeting of the ORF3 protein (14, 20). The cations for other plant and animal viruses that interact with the whole L-rich region also acts as a nuclear export signal (20), nucleolus. suggesting that the ORF3 protein traffics between the nucleus (nucleolus) and cytoplasm of infected cells. Cajal bodies ͉ plant virus movement ͉ ribonucleoprotein particles The ORF3 protein is produced in the cytoplasm, enters the nucleus, and is targeted to CBs. The CBs are then reorganized into he nucleolus is a subnuclear domain and is the site of transcrip- multiple smaller structures (CB-like aggregates, CBLs) that move Ttion and processing of rRNA and of ribosome biogenesis. In to and fuse with the nucleolus by an unknown mechanism (14). The addition, the nucleolus also participates in other aspects of RNA ORF3 protein is exported from the nucleus, leading to the forma- metabolism and cell function (1, 2). The nucleolus is structurally tion of cytoplasmic viral RNP particles that are transported to the and functionally associated with Cajal bodies (CBs), which are rest of the plant via the phloem. The integral connection between structures found in both animals and plants (3, 4). CBs contain nucleolar targeting of the ORF3 protein and its biological function different proteins including coilin, a protein essential for CB in virus long-distance spread has been demonstrated by the intro- formation, and fibrillarin, a major nucleolar protein that is a core duction of mutations in the R- and L-rich domains that block component of small nucleolar ribonucleoprotein particles nucleolar localization and nuclear export of the ORF3 protein, (snoRNPs) and is required for rRNA processing (4–7). CBs are involved in the maturation of small nuclear RNPs (snRNPs) and snoRNPs, which traffic through CBs before accumulating in splic- Author contributions: S.H.K., S.M., N.O.K., J.W.S.B., and M.T. designed research; S.H.K., ing speckles and the nucleolus, respectively (8, 9). Both the nucle- S.M., D.V.R., E.V.R., T.G., and S.H. performed research; S.H.K., N.O.K., D.V.R., J.W.S.B., and M.T. analyzed data; and S.H.K., J.W.S.B., and M.T. wrote the paper. olus and CBs have a role in RNA silencing in plants (10, 11). Finally, The authors declare no conflict of interest. a number of animal and plant viruses including the RNA- Abbreviations: CB, Cajal bodies; CBL, CB-like aggregates; GRV, groundnut rosette virus; containing tobacco etch virus and the DNA-containing tomato TMV, tobacco mosaic virus; PVX, potato virus X; TRV, tobacco rattle virus; RNP, ribonucle- yellow leaf curl virus have a nucleolar phase in their life cycle (12, oprotein; Fib2, fibrillarin 2; GAR, glycine- and arginine-rich domain; CP, coat protein. 13). Recently, we have shown that the ability of the umbravirus, Data deposition: The nucleotide sequence of the NbFib cDNA was deposited in GenBank groundnut rosette virus (GRV), to move long distances through the (accession no. AM269909). phloem, the specialized vascular system used by plants for the ¶To whom correspondence should be addressed. E-mail: [email protected]. transport of assimilates and macromolecules, depends strictly on This article contains supporting information online at www.pnas.org/cgi/content/full/ the interaction of one of its proteins, the ORF3 protein, with CBs 0704632104/DC1. and the nucleolus (14). © 2007 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0704632104 PNAS ͉ June 26, 2007 ͉ vol. 104 ͉ no. 26 ͉ 11115–11120 Downloaded by guest on September 24, 2021 Fig. 1. Correlation between the ability of the ORF3 protein to traffic through the nucleolus and the relocalization of fibrillarin, formation of viral RNPs, and long-distance movement. Wild-type and mutant ORF3 protein sequences of the R-rich and L–rich domains are shown in combination with data on nuclear (N) and nucleolar (No) localization, nuclear export (N-exp) of the ORF3 protein, relocalization of fibrillarin (Cyt. fibrillarin), RNP formation, and virus long-distance movement (LDM) (14).

respectively, resulting in failure to form viral RNPs, and of their long-distance movement (14) (Fig. 1). In elucidating the nuclear pathway of the ORF3 protein, we also observed the partial relocalization of the nucleolar protein, fibrillarin, to the cytoplasmic inclusions containing viral RNPs, whereas normally, fibrillarin does not accumulate in cytoplasm (14). Fibril- larin is one of the major proteins of the nucleolus. It is also localized to CBs, is a core component of box C/D snoRNPs, and has Fig. 2. Virus-induced silencing of the fibrillarin gene (NbFib) in N. benthami- methyltransferase activity directing methylation of rRNA and snR- ana plants by using a TRV vector. (A–C) Expression of fibrillarin was suppressed NAs (21). Interaction of some animal viruses with fibrillarin and by TRV-NbFib to different levels, and plants with effectively three different other nucleolar proteins has been reported (12, 13, 22, 23), but their phenotypes (I, II, and III) were generated. (A) Growth phenotypes of silenced specific role in virus infections remains elusive. In this article, we plants in comparison with control (c) nonsilenced plants. (B) Semiquantitative demonstrate a previously uncharacterized function of fibrillarin in RT-PCR analysis of NbFib mRNA accumulation. Ethidium bromide-stained umbravirus systemic infection. The GRV ORF3 protein directly agarose gels show RT-PCR products corresponding to the fragments of NbFib interacts with fibrillarin, and this interaction is essential for nucle- (Fib) mRNA (320 bp) and ubiquitin mRNA (176 bp) used as a control, as indicated by arrows. Lanes 1–3 represent plant replicates. (C) Western blot olar localization of the ORF3 protein. Fibrillarin is also required, analysis of fibrillarin (Fib) accumulation. The position of fibrillarin is shown by along with umbravirus ORF3 protein and viral RNA, for the an arrow. Lanes 1–3 as above. (D) Immunofluorescence staining of cells of assembly of movement-competent, infectious RNP particles. Thus, fibrillarin-silenced (group II; Fib-s) and nonsilenced (Non-s) plants with fibril- the ORF3 protein may exploit fibrillarin trafficking to reach the larin and coilin antibodies visualized by confocal microscopy. Fibrillarin levels nucleolus and use the properties of fibrillarin to form umbraviral are significantly reduced in the nucleolus and are undetectable in CBs, visu- RNPs. alized by using antibody against coilin. (Scale bars, 5 ␮m.) Results Fibrillarin Knockdown Suppresses Long-Distance Movement of GRV. partially silenced plants had a normal growth phenotype (Fig. 2A) Although fibrillarin is required for essential cell functions, a sig- and were able to support infection by tobacco mosaic virus (TMV) nificant (Ϸ50%) decrease in levels of fibrillarin is tolerated in mice and potato virus X (PVX), accumulating virus in both inoculated (24). We therefore examined the effect of silencing of fibrillarin and uninoculated systemically infected leaves to high levels (Table expression in plants on the nucleolar localization of the ORF3 1). This suggests that reduction of fibrillarin levels in the knockdown protein and its function in virus long-distance movement. A virus- plants (in both nucleoli and CBs) did not inhibit normal cellular induced gene-silencing construct containing a fragment of the functions, such as translation, that are necessary for general virus fibrillarin gene from Nicotiana benthamiana was made by using a infection. Tobacco rattle virus (TRV) vector, giving TRV-NbFib [supporting In contrast to TMV and PVX, GRV was unable to cause systemic information (SI) Text]. As a nonsilenced control, N. benthamiana plants were infected with TRV alone. Expression of fibrillarin mRNA and production of fibrillarin were suppressed by TRV- Table 1. Accumulation of TMV and PVX in inoculated and NbFib to different levels, and plants with effectively three different uninoculated leaves determined by ELISA and phenotypes were generated: (Group I) dwarf plants with necrosis by infectivity assays showing complete or very strong silencing (Ͼ90% decrease in the Fib-silenced (II) Nonsilenced expression of fibrillarin), (Group II) symptomless plants showing a partial (50–90%) reduction of fibrillarin expression, and (Group Virus Leaf ELISA Infectivity ELISA Infectivity III) symptomless plants showing no or very weak (Ͻ10%) reduction TMX In 1.2 Ϯ 0.2 55 Ϯ 4 1.1 Ϯ 0.1 62 Ϯ 8 of fibrillarin expression (Fig. 2 A–C). Because of the severely U 1.3 Ϯ 0.1 78 Ϯ 7 1.2 Ϯ 0.2 59 Ϯ 7 affected growth or the low level of silencing of the first and third PVX In 0.8 Ϯ 0.1 41 Ϯ 5 0.6 Ϯ 0.1 38 Ϯ 6 classes of plants, respectively, plants of the second type were U 1.1 Ϯ 0.1 64 Ϯ 7 1.0 Ϯ 0.2 41 Ϯ 5 selected for further analysis. Immunofluorescence analysis of these Ϯ plants by using anti-fibrillarin antibody showed that fibrillarin levels ELISA data are means (A405) SD (for details see SI Text). Infectivity was were considerably reduced in the nucleolus and were undetectable determined by inoculation of virus samples extracted from inoculated (In) and uninoculated (U) leaves on half leaves of N. tabacum “Xanthi-nc” (TMV) and in CBs in any of the 50 analyzed cells (Fig. 2D). CBs were visualized C. amaranticolor (PVX) and is presented as the number of local lesions per in silenced and nonsilenced plants by using antibody against coilin, half-leaf. Values are means Ϯ SD (for details, see SI Text). To eliminate TRV a standard protein marker of CBs, which detected usually one but infectivity, the samples were pretreated with antibody against TRV as de- up to three CBs per cell, as normal for these cells (Fig. 2D). These scribed in SI Text.

11116 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0704632104 Kim et al. Downloaded by guest on September 24, 2021 move systemically, showing a correlation between the amount of fibrillarin expressed in plants and the ability of GRV to move long distances (Table 2).

ORF3 Protein Does Not Localize to the Nucleolus in Fibrillarin Knock- down Plants. We next examined the subnuclear localization of the ORF3 protein using a fusion with green fluorescent protein (ORF3-GFP) delivered by Agrobacterium into cells of fibrillarin- silenced plants (Group II). The ORF3 protein localized to the nucleolus of nonsilenced plants as expected but did not localize to the nucleolus of silenced plants (Fig. 3C). Instead, the ORF3-GFP protein localized to multiple subnuclear structures (Fig. 3C). These contained coilin (data not shown) and were similar to the multiple CBLs (CB-like structures) observed previously upon GRV infection or with transient expression of ORF3 (14). This cellular phenotype mimics that seen with the ORF3 protein mutant, where the single leucine L149 was replaced with alanine (L149A-GFP; ref. 14; see Fig. 1). This mutant is defective in viral long-distance movement and is unable to locate to the nucleolus but, instead, accumulates in multiple CBLs (14) (Fig. 1). These results further confirm the correlation between localization of the ORF3 protein to the nucleolus and the ability of the virus to move systemically through the plant (14) and suggest that fibrillarin is involved in the fusion of CBs with the nucleolus and, thereby, the trafficking of the ORF3 protein to the nucleolus. Fig. 3. Fibrillarin knockdown suppresses long-distance movement of GRV and nucleolar localization of ORF3 protein. (A and B) Accumulation of GRV-YB Interaction of ORF3 Protein with Fibrillarin. The relocalization of in inoculated (in) and uninoculated (u) leaves of nonsilenced (Non-s) and some fibrillarin to cytoplasmic inclusions along with the ORF3 ﬁbrillarin-silenced (Fib-s) N. benthamiana plants of group II (A, B, and Table 2). protein (14) and the altered ORF3 protein localization and its lack (A) Symptoms on systemically infected, uninoculated leaves. Nonsilenced of function in virus long-distance movement in fibrillarin-silenced plants show yellow blotch symptoms characteristic of the GRV-YB isolate, plants suggest that fibrillarin is essential for the GRV infection whereas ﬁbrillarin-silenced plants (group II) are symptomless. (B) Northern process. To determine whether there was a direct interaction blot analysis of RNA isolated from leaves of the infected plants. GRV RNA is shown by an arrow; rRNA bands were stained by ethidium bromide for loading between fibrillarin and the ORF3 protein, we examined the in vitro control (lower blot). (C) Localization of the ORF3-GFP protein delivered by binding of recombinant GST-tagged fibrillarin [Arabidopsis fibril- Agrobacterium into cells of the nonsilenced and silenced plants. No, nucleo- larin 2 (Fib2), Fig. 4A] (25) by the ORF3 protein. The ORF3

lus; CBL, Cajal body-like structures; N, nuclei (shown by dashed lines according protein was expressed in plants from a TMV vector, isolated, and PLANT BIOLOGY to DAPI staining). (Scale bars, 5 ␮m.) tested by using a blot overlay (far Western) assay. Fibrillarin- and GST-containing proteins were separated by SDS/PAGE, transferred to nitrocellulose membrane, renatured, incubated with pu- infection in the fibrillarin-silenced plants. Uninoculated tip leaves rified ORF3 protein, and protein binding was visualized by using of these plants did not develop infection symptoms (Fig. 3A). anti-ORF3 protein antibodies. For Fib2, two protein bands of 60 Accumulation of GRV in the inoculated leaves was comparable and 30–35 kDa were bound by the ORF3 protein (Fig. 4B, lane 1). with that in inoculated leaves of nonsilenced plants, as shown by the The larger band corresponded to GST-fibrillarin whereas the presence of GRV RNA (Fig. 3B) and by infectivity tests (Table 2), smaller protein was a fragment of fibrillarin (Fib2*) containing the but upper uninoculated leaves of the silenced plants did not contain N-terminal 79 amino acid residues still tagged with GST, as GRV RNA or infectious viral RNP (Fig. 3 B and C). Thus, the determined by mass spectroscopy (data not shown; Fig. 4A). The fibrillarin deficiency in these silenced plants did not affect the ability GST-Fib2* fragment is presumably produced by degradation of the of the virus to replicate and move locally from cell to cell but protein or truncation during translation. GST alone did not interact inhibited long-distance virus movement. In agreement with this, in with the ORF3 protein antibody (Fig. 4B, lane 3). Removal of GST from the recombinant fibrillarin by thrombin treatment did not the partially silenced Group III plants that continue to express disrupt binding of the ORF3 protein (Fig. 4B, lane 4). The substantial amounts of fibrillarin (Fig. 2 A–C), GRV was able to N-terminal fragment of fibrillarin contains the glycine- and arginine-rich (GAR) domain (Fig. 4A) (25, 26) responsible for interaction with other proteins such as spinal muscular atrophy disease Table 2. Accumulation of GRV-YB in inoculated (In) and protein (27). A deletion mutant, where the GAR domain was uninoculated (U) leaves of nonsilenced (Non-s) and removed from GST-fibrillarin (Fig. 4A; GST-Fib2⌬GAR), was ﬁbrillarin-silenced (Fib-s) N. benthamiana plants of Group III unable to interact with the ORF3 protein (Fig. 4B, lane 2). Taken Infectivity together, these results demonstrate that the ORF3 protein directly interacts with fibrillarin in vitro and that the N-terminal GAR GRV Non-s Fib-s (II) Fib-s (III) domain of fibrillarin is necessary for this interaction. In 25 Ϯ 434Ϯ 628Ϯ 2 To determine the effect on the interaction of the ORF3 protein U35Ϯ 70 31Ϯ 5 and fibrillarin of mutations in the conserved R- and L-rich domains involved in ORF3 protein nuclear trafficking (Fig. 1) mutant Infectivity of extracts from inoculated (In) and uninoculated (U) leaves determined by inoculation on half-leaves of C. quinoa plants and presented as proteins were produced in plants after expression from TMV the number of local lesions per half-leaf. Half-leaves were inoculated with an vectors. The mutations are shown in Fig. 1 and have been described amount of extract equivalent to 1 mg of tissue. Values are means Ϯ SD (for previously (14). Briefly, replacement of all six arginine residues in details, see SI Text). To eliminate TRV infectivity, the samples were pretreated the R-rich domain by alanine residues gave the RA protein (Fig. 1). with antibody against TRV. Mutations to the L-rich region replaced either all of the invariant

Kim et al. PNAS ͉ June 26, 2007 ͉ vol. 104 ͉ no. 26 ͉ 11117 Downloaded by guest on September 24, 2021 Fig. 5. EM images showing complexes generated by different mixtures of umbraviral RNA, ORF3 protein, and fibrillarin. (A) Complex of GRV ORF3 protein and PEMV-2 RNA (F, thin filaments; DC, densely coated RNA). (B) Complex of Arabidopsis fibrillarin (Fib2) and PEMV-2 RNA (F and DC as above). (C) Disk-like complexes formed by ORF3 protein and Fib2 in the absence of RNA. (D) Complexes of ORF3 protein, Fib2 and RNA (an arrow shows elements of helical structure). Insets show higher magnification. (Scale bars: A–D, 100 nm; Insets, 50 nm.)

protein were mixed at a 1:400 molar ratio (previously shown to saturate RNA binding with the ORF3 protein) (16). In EM, this Fig. 4. Fibrillarin interacts with ORF3 protein. (A) Representation of Arabi- ␣ gave nonuniform structures (Fig. 5A) that were clearly distinct from dopsis fibrillarin 2 (Fib2) showing GAR, RNA binding and -helix domains, the RNP complexes formed by the ORF3 protein in vivo (16). The GST fusion protein (GST-Fib2), GST fused with Fib2 fragment (GST-Fib2*) and ⌬ ORF3 protein–viral RNA complexes consisted of two structural the GAR deletion mutant (GST-Fib2 GAR). Numbers are amino acid residue Ϸ positions; triangle shows position of thrombin cleavage site. (B) Far Western elements: (i) thin interwoven filaments with a diameter of 3–4 nm blot analysis of the interaction between ORF3 protein and fibrillarin, Fib2. and (ii) thicker (Ϸ6–8 nm diameter) filaments, possibly reflecting Blots containing GST-Fib2 (also containing GST-Fib2*; lane 1), GST-Fib2⌬GAR different densities of protein coating the RNA (Fig. 5A). Increasing (lane 2), GST (lane 3) and Fib2 (GST-Fib2 treated with thrombin; lane 4) were the relative amount of ORF3 protein did not affect the structure of incubated with or without the ORF3 protein [ϩORF3 (Left) and ϪORF3 (Cen- the complexes. Viral RNA and fibrillarin (Fib2) (at molar ratios of ter)] and probed by anti-ORF3 antibody. (Right) Coomassie staining. Positions 1:400 to 1:800) formed complexes similar to those formed by the of GST-Fib2, GST-Fib2*, and Fib2 are shown on the left and those of molecular ORF3 protein and viral RNA (Fig. 5B). RNA alone did not form mass markers are on the right. (C) Far Western blot analysis of the interaction between the purified recombinant ORF3 protein mutants and fibrillarin. Blots any visible filaments in the same EM procedure (data not shown). containing GST-Fib2 (and GST-Fib2*) were incubated with wild-type (1) or Mixing of the ORF3 protein and fibrillarin in the absence of viral mutant ORF3 proteins (LA, 2; L149A, 3; L153A, 4; RA, 5) and probed as above. RNA also did not give rise to filamentous structures but led to Lane 7 represents Ponceau S staining of fibrillarin. Lane 6 corresponds to a formation of disk-like complexes with a diameter of Ϸ18–22 nm control sample prepared from a plant infected with the TMV vector alone by (Fig. 5C) possibly consisting of both ORF3 protein and fibrillarin using the same protocol as for the recombinant ORF3 proteins. The lower blot molecules. shows expression levels of wild-type and mutant ORF3 proteins (Coomassie In contrast, incubation of viral RNA with both the ORF3 protein staining) as indicated. M, marker of 30 kDa. and fibrillarin (at molar ratios of 1:400:400) led to the formation of regular filamentous structures (Fig. 5D) similar to viral RNP particles formed in vivo after GRV infection. In EM, the particles leucine residues (positions 149, 152, and 153, LA), or individual had a diameter of Ϸ20 nm, and helical repeat structures were leucines, with alanine residues (Fig. 1). Of these mutants, only LA observed on some filaments (Fig. 5D), which is in good agreement and L149A were unable to bind fibrillarin in the far Western assay with dimensions and structure of RNP particles formed by the GRV (Fig. 4C), indicating that the L-rich domain (and L149 in particular) ORF3 protein in vivo. No regular RNP particles were formed with is involved in the interaction with fibrillarin. the fibrillarin mutant lacking the GAR domain (Fib2⌬GAR), which does not interact with the ORF3 protein (data not shown). Fibrillarin Mediates Assembly of Umbravirus RNP Particles in Vitro. Thus, the interaction between fibrillarin and the ORF3 protein is The ORF3 protein binds to umbraviral RNA in vitro but forms necessary to mediate formation of RNP particles similar in struc- RNA–protein complexes with a completely different structure from ture to GRV RNP particles formed in vivo. viral RNPs in infected cells (16), suggesting that other factor(s) are involved in viral RNP assembly in vivo. Because of the ORF3 Viral RNP Particles Formed in Vitro Are Infectious. In addition to protein–fibrillarin interaction and the presence of fibrillarin in structural similarities, the RNP particles formed from the ORF3 umbraviral cytoplasmic inclusions, we examined the effect of protein, fibrillarin, and viral RNA in vitro were infectious and fibrillarin on the interaction of the ORF3 protein with viral RNA induced numerous lesions on the test plant, Chenopodium quinoa using EM. Umbraviral RNA, transcribed from a full-length cDNA (Table 3). The infectivity of the in vitro RNP particles was neutral- clone of pea enation mosaic virus-2 (PEMV-2), an umbravirus ized by both anti-ORF3 protein and anti-fibrillarin antibodies, closely related to GRV (15), was added to fibrillarin, to the ORF3 whereas control anti-TMV antibody did not affect the infectivity protein, or to mixtures of both proteins. Viral RNA and ORF3 (Table 3), suggesting that both (ORF3 and fibrillarin) proteins are

11118 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0704632104 Kim et al. Downloaded by guest on September 24, 2021 Table 3. Infectivity of RNP complexes formed by the ORF3 could be due to ‘‘off-target’’ silencing events, taken together with protein, fibrillarin, and PEMV-2 RNA results on the inability of the ORF3 L149A and LA mutant proteins RNP components Treatment Lesions* to interact with fibrillarin (Fig. 4C), our data strongly support the idea that fibrillarin is essential for both the fusion of CBLs with the PEMV-2 RNA, none, fibrillarin RNase 0 Ϯ 0 nucleolus and for umbravirus systemic infection. Ϯ PEMV-2 RNA, ORF3 protein, none RNase 2 1 The physical association of the nucleolus and CBs is well docu- Ϯ PEMV-2 RNA, ORF3 protein, fibrillarin None 63 6 mented and is controlled by molecular interactions among CB and Ϯ PEMV-2 RNA, ORF3 protein, fibrillarin RNase 72 11 nucleolar proteins (4, 6). The interaction of the ORF3 protein with PEMV-2 RNA, ORF3 protein, fibrillarin Ab-Fib 5 Ϯ 1 fibrillarin may interfere with normal host protein–protein interac- PEMV-2 RNA, ORF3 protein, fibrillarin Ab-ORF3 3 Ϯ 1 tions or other processes affecting CB integrity, which, in turn, may PEMV-2 RNA, ORF3 protein, fibrillarin Ab-TMV 62 Ϯ 7 cause the fusion of CBLs with the nucleolus. Given that the ORF3 *Number of lesions was determined by inoculation of the untreated or protein and fibrillarin are able to interact and that both move to the treated RNA–protein complexes with RNase A (15 ␮g/ml), anti-fibrillarin nucleolus via CBLs, it is conceivable that they may move as a (Ab-Fib), anti-ORF3 protein (Ab-ORF3), or control anti-TMV antibodies (Ab- complex. However, it is also possible that some of the ORF3 protein TMV) on half-leaves of C. quinod. Values are means of triplicate samples Ϯ SD (for details, see SI Text). The amounts of complexes inoculated were pool enters the nucleolus via CBLs as uncomplexed molecules, equivalent to 25 ng of viral RNA per half-leaf. where they can interact with fibrillarin to form ORF3-fibrillarin complexes for export to cytoplasm. Previously, we have shown that umbraviral RNP assembly occurs present in the RNP particles. Complexes formed by the ORF3 in the cytoplasm (16). Fibrillarin is needed for cytoplasmic RNP protein and viral RNA in the absence of fibrillarin were not formation. The redistribution of fibrillarin with the ORF3 protein, infectious (Table 3). Finally, infectivity of the RNP particles was not therefore, is a prerequisite for formation of umbraviral RNP affected by RNase treatment (Table 3), showing protection of the particles. Here, we demonstrate in vitro that fibrillarin, in combi- viral RNA by the ORF3 protein and fibrillarin. nation with the ORF3 protein and viral RNA produces filamentous RNP particles with elements of helical architecture and structural Discussion properties similar to the viral RNPs formed in vivo described by To mediate virus long-distance movement and systemic infection, Taliansky et al. (16). Moreover, these RNP particles were infectious the umbravirus-encoded ORF3 protein is imported from the and protected the viral RNA from RNase treatment. This suggests cytoplasm into the nucleus and nucleolus and then returns to the that the ORF3 protein and fibrillarin are sufficient for encapsida- cytoplasm. The GRV ORF3 protein traffics to the nucleolus by a tion of the viral RNA into an infectious RNP filament capable of process involving the reorganization of CBs into CBLs and their fusion with the nucleolus. The most likely purpose of this nuclear/ long-distance movement. nucleolar trafficking of the ORF3 protein is to recruit the major Fibrillarin, an RNA-binding protein, may bind the viral RNA or nucleolar protein, fibrillarin, and relocalize it to cytoplasm (14). may act to permit or catalyze the regular assembly of proteins Here, we show that the GRV ORF3 protein is able physically to around viral RNA, which is unachievable by the ORF3 protein alone. When formed in phloem companion cells, the viral RNPs interact with fibrillarin. Our studies also demonstrate that fibrillarin PLANT BIOLOGY is involved in at least two stages in the GRV life cycle: (i) CBL presumably can enter sieve elements and move through the plant to fusion with the nucleolus is mediated by the interaction between the cause systemic infection. These RNP complexes may also be ORF3 protein and fibrillarin, and (ii) the ORF3 protein redistrib- involved in protecting viral RNA from the plant’s defensive RNA utes some fibrillarin to the cytoplasm, where it functions in the silencing response mediated by short interfering RNAs, sustaining formation of viral RNP particles that are capable of long-distance the viral life cycle. Whether the disk-shaped particles formed by the movement and causing systemic viral infection. This model dem- ORF3 protein and fibrillarin are required for RNP formation and onstrates previously uncharacterized functions for fibrillarin in how they interact with viral RNA to organize the helical structure triggering nucleolar import of the ORF3 protein by CBLs by an remains to be determined. Nevertheless, the interaction of the unknown mechanism and mediating assembly of umbraviral RNPs. GRV ORF3 protein with fibrillarin in CBs/CBLs triggers all of the The trafficking of the ORF3 protein to CBs and their reorgani- molecular and cellular events necessary to establish a systemic zation into CBLs does not require the ORF3–fibrillarin interaction, infection. because the L149A mutant, which is unable to interact with Because umbraviruses lack a CP, the main functions of CPs of fibrillarin, enters the nucleus and accumulates in CBLs (14). In other viruses (formation of virus particles, systemic spread, and contrast, the fusion of CBLs with the nucleolus is presumably aphid transmission) must be replaced somehow. The ORF3 pro- mediated by the direct interaction between the L-rich region (L149 tein–fibrillarin interaction evidently takes care of the RNP forma- in particular) of the ORF3 protein and the GAR domain of tion and long-distance movement functions. For aphid transmis- fibrillarin. This interaction is likely to occur in CBs or CBLs because sion, umbraviral RNA is packaged in the CP of an unrelated helper the L149A mutant does not interact with fibrillarin or cause CBL virus that is itself aphid-transmitted (15). Thus, key functions of fusion with the nucleolus. Further support of a role for fibrillarin in viral CP are, in umbraviruses, replaced by these two different this fusion is provided by the fibrillarin knockdown experiments. mechanisms. When no fibrillarin was detectable in CBs/CBLs, the interaction between the ORF3 protein and fibrillarin could not occur, and the We have recently shown that the CP of another plant virus, ORF3 protein accumulated in multiple CBLs that did not fuse with potato leafroll virus (PLRV, a polerovirus), targets the nucleolus the nucleolus. Thus, the fibrillarin-silenced plants (Group II) ex- (28). Our preliminary results also indicate that, like GRV, PLRV hibited a phenotype equivalent to that of plants infected with the is unable to cause systemic infection in the fibrillarin-silenced L149A or LA umbravirus ORF3 mutants: neither mutant allowed Group II plants, although accumulation of PLRV in the inoculated the fusion of CBLs with the nucleolus or systemic spread (14). It is leaves is not affected (SI Table 4). This suggests that fibrillarin also important that both TMV and PVX were able to move long involvement in virus infection could be a more general mechanism distances in fibrillarin-silenced plants, suggesting that fibrillarin is among viruses with a nucleolar phase of life cycle. Finally, some involved only in the movement of plant viruses, such as umbravi- RNA silencing pathways also involve the nucleolus and CBs (10, 11) ruses, that have a nucleolar phase of infection. Although it is and, hence, may be similar to the virus movement pathway de- possible that the effects seen in the knockdown (Group II) plants scribed here.

Kim et al. PNAS ͉ June 26, 2007 ͉ vol. 104 ͉ no. 26 ͉ 11119 Downloaded by guest on September 24, 2021 Materials and Methods 488-conjugated anti-mouse or anti-rabbit secondary antibodies Virus Strains and TMV-Based Constructs. GRV-YB was maintained as (Invitrogen, Carlsbad, CA). a stock isolate. The PEMV-2 cDNA clone was described by Ryabov et al. (29). TMV(ORF3-His) expressing His-tagged ORF3 protein Expression and Purification of the ORF3 Protein, Fibrillarin, and was described (16). Changes were introduced into ORF3 coding Mutants of Each. ORF3-His protein and its mutants were expressed sequences of the construct TMV(ORF3-His) to produce mutants in N. benthamiana from TMV and purified by Ni2ϩ-affinity chro- listed in Fig. 1 by PCR using self-complementary mutagenic matography (16). Arabidopsis fibrillarin 2 (Fib2) coding sequence primers. Preparation of in vitro RNA transcripts and inoculation of was cloned into pGEX-KG (32) to generate a construct with Fib2 plants were as described (17). tagged at the N terminus with GST (GST-Fib2). A deletion mutant where the N-terminal 77 amino acid residues (GAR domain; Fig. Analysis of Fibrillarin-Silenced Plants. RT-PCR analysis of the accu- 4A) were removed (GST-Fib2⌬GAR) was produced by PCR using mulation of fibrillarin mRNA (NbFib) was performed by using self-complementary mutagenic primers. The GST-fused proteins primers corresponding to a 320-nt fragment (SI Text). Amplifica- were expressed in Escherichia coli and purified by glutathione- tion of a fragment of ubiquitin mRNA (176 nt) was used as a agarose chromatography, and GST was removed by thrombin control. For Western blot analysis, proteins were fractionated on treatment as described (32). 10% SDS/PAGE and transferred onto nitrocellulose membrane. The blots were probed by anti-fibrillarin antibodies. The immune Far Western Blot Analysis of the ORF3 Protein–Fibrillarin Interaction. reactions were visualized by ECL system (Amersham Biosciences, ⌬ Uppsala, Sweden). ELISA of TMV and PVX accumulation was GST-Fib2, GST-Fib2 GAR, GST, or Fib2 were fractionated on carried out essentially as described (30) (for details, see SI Text). 10% SDS/PAGE and transferred onto nitrocellulose membrane Virus infectivity in samples from inoculated and uninoculated (Roche Applied Science, Indianapolis, IN). The blots were rena- leaves infected with TMV, PVX, and GRV was assessed by tured, blocked, incubated with the ORF3 protein or its mutants (5 counting the mean number of lesions induced in half leaves of N. ␮g/ml), and probed with anti-ORF3 antibody essentially as de- tabacum‘‘Xanthi-nc’’ (TMV) and C. quinoa (PVX and GRV) scribed (22). The reaction was visualized by using the ECL system. plants. For Northern blot analysis of GRV RNA, RNA extracted from plants was probed with 32P-labeled cDNA probe correspond- Formation of ORF3 RNP Complexes in the Presence of Fibrillarin. RNA ing to the clone gr21GRV (31). transcribed from a cDNA clone of PEMV-2 (29) (75 ng) was heat-denatured at 90°C for 1 min and mixed with saturating Generation and Agrobacterium-Mediated Expression of ORF3-GFP amounts of the ORF3 protein and fibrillarin (as indicated in Construct. The ORF3 protein fused to the N terminus of GFP Results)in15␮l of buffer A [10 mM Hepes-KOH (pH 7.6)/100 mM (ORF3-GFP) was generated by using the Gateway recombination KCl]. For EM, the samples were stained with 2% sodium phos- system and delivered by Agrobacterium into N. benthamiana leaves photungstate and examined in a CM 10 electron microscope (Royal as described (14). Philips Electronics, Amsterdam, The Netherlands).

Confocal Imaging Analysis. Localization of GFP fused to ORF3 We thank P. Shaw (John Innes Centre, Norwich, U.K.) for providing protein was monitored by using a TCS SP2 (Leica Microsystems, antibodies to plant coilin. This work was supported by a grant-in-aid from Wetzlar, Germany) confocal laser scanning microscope. To locate the Scottish Executive Environment and Rural Affairs Department, by nuclei, the leaf tissues were infiltrated with PBS containing 4Ј,6Ј- the Royal Society (to J.W.S.B., N.O.K., and M.T.), Biotechnology and diamidino-2-phenylindole (DAPI). Coilin and fibrillarin were lo- Biological Sciences Research Council (E.V.R.), International Associa- calized by using primary rabbit antibody to Atcoilin and primary tion for the Promotion of Cooperation with Scientists from the New monoclonal antibody to human fibrillarin (Cytoskeleton, Denver, Independent States of the former Soviet Union (INTAS) Fellowship (to CO). The primary antibodies were visualized by Alexa Fluor D.V.R.), and the Russian Foundation for Basic Research (N.O.K.).

1. Rubbi CP, Milner J (2003) EMBO J 22:6068–6077. 18. Ryabov EV, Robinson DJ, Taliansky ME (2001b) Virology 288:391–400. 2. Olsen MOJ (2004) In The Nucleolus, ed Olsen MOJ (Landes, Georgetown, 19. Ryabov EV, Oparka KJ, Santa Cruz S, Robinson DJ, Taliansky ME (1998) TX/Kluwer, New York), pp 329–342. Virology 242:303–313. 3. Beven AF, Simpson GG, Brown JWS, Shaw PJ (1995) J Cell Sci 108:509–518. 20. Ryabov EV, Kim SH, Taliansky ME (2004) J Gen Virol 85:1329–1333. 4. Cioce M, Lamond AI (2005) Annu Rev Cell Dev Biol 21:105–131. 21. Filipowicz W, Pogae`ic´ V (2002) Curr Opin Cell Biol 14:319–327. 5. Venema J, Tollervey D (1999) Annu Rev Genet 33:261–311. 22. Chen MH, Sheng JS, Hind G, Handa AK, Citovsky V (2000) EMBO J 6. Ogg SC, Lamond AI (2002) J Cell Biol 159:17–21. 19:913–920. 7. Matera AG, Shpargel KB (2006) Curr Opin Cell Biol 18:317–324. 23. Yoo D, Wootton SK, Li G, Song C, Rowland RR (2003) J Virol 8. Sleeman JE, Lamond AI (1999) Curr Biol 9:1065–1074. 77:12173–12183. 9. Sleeman JE, Ajuh P, Lamond AI (2001) J Cell Sci 114:4407–4419. 24. Newton K, Petfalski E, Tollervey D, Caceres JF (2003) Mol Cell Biol 23:8519– 10. Li CF, Pontes O, El-Shami M, Henderson IR, Bernatavichute YV, Chan SW-L, 8527. Lagrange T, Pikaard CS, Jacobsen SE (2006) Cell 126:93–106. 25. Barneche F, Steinmetz F, Echeverria M (2000) J Biol Chem 275:27212– 11. Pontes O, Li CF, Nunes PC, Haag J, Ream T, Vitins A, Jacobsen SE, Pikaard 27220. CS (2006) Cell 126:79–92. 26. Aris JP, Blobel G (1991) Proc Natl Acad Sci USA 88:931–935. 12. Hiscox JA (2002) Arch Virol 147:1077–1089. 27. Jones KW, Gorzynski K, Hales CM, Fischer U, Badbanchi F, Terns RM, Terns 13. Hiscox JA (2007) Nat Rev Microbiol 5:119–127. MP (2001) J Biol Chem 276:38645–38651. 14. Kim SH, Ryabov EV, Kalinina NO, Rakitina DV, Gillespie T, MacFarlane S, 28. Haupt S, Stroganova T, Ryabov E, Kim SH, Fraser G, Duncan G, Mayo MA, Haupt S, Brown JWS, Taliansky M (2007) EMBO J 26:2169–2179. Barker H, Taliansky M (2005) J Gen Virol 86:2891–2896. 15. Taliansky ME, Robinson DJ (2003) J Gen Virol 84:1951–1960. 29. Ryabov EV, Fraser G, Mayo MA, Barker H, Taliansky M (2001) Virology 16. Taliansky ME, Roberts IM, Kalinina N, Ryabov EV, Raj SK, Robinson DJ, 286:363–372. Oparka KJ (2003) J Virol 77:3031–3040. 30. Barker H, Solomon RM (1990) Theoret Appl Genet 80:188–192. 17. Ryabov EV, Robinson DJ, Taliansky ME (1999) Proc Natl Acad Sci USA 31. Taliansky ME, Robinson DJ, Murant AF (1996) J Gen Virol 77:2335–2345. 96:1212–1217. 32. Guan KL, Dixon JE (1991) Anal Biochem 192:262–267.

11120 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0704632104 Kim et al. Downloaded by guest on September 24, 2021