![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Previous Article | Next Article 
Journal of Virology, October 2004, p. 11161-11171, Vol. 78, No. 20
0022-538X/04/$08.00+0 DOI: 10.1128/JVI.78.20.11161-11171.2004
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Department of Plant Pathology, Cornell University, Ithaca, New York
Received 10 April 2004/ Accepted 2 June 2004
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although it is well recognized that movement proteins facilitate cell-to-cell transport of the viral genome by modifying plasmodesmata (37), the complex transwall channels that connect adjacent plant cells, little is known about how movement proteins reach plasmodesmata or the host factors that regulate movement protein function. Subcellular localization studies have shown that SqLCV and CLCV MPs target to the cortical endoplasmic reticulum (ER) (47, 57; J. D. Lewis and S. G. Lazarowitz, unpublished data) and that the Tobacco mosaic virus (TMV) 30-kDa movement protein can associate with microtubules and actin filaments as well as with the ER (19, 31, 33). Gillespie and colleagues (16), in a recent study on the accelerated movement of a TMV mutant encoding a functionally enhanced 30-kDa movement protein [MP(R3)], reported that in contrast to the prevailing view that the interaction of the 30-kDa protein with microtubules is crucial for cell-to-cell transport of the viral RNA genome (19, 31, 33), movement protein interactions with the cortical ER are important and cell-to-cell trafficking of TMV can occur independently of microtubules. The interaction with microtubules appears to downregulate the levels of TMV movement protein, acting to draw the 30-kDa protein away from the ER and target it for proteasome degradation via a well-recognized cellular pathway (16). Consistent with this, Kragler and colleagues (26) found that the TMV 30-kDa protein interacts with a microtubule-associated protein which does not seem to be required for movement but does appear to negatively regulate the process.
A number of groups have used protein-protein interaction assays to identify host proteins that are potentially relevant to virus movement. Using a gel-overlay assay or a yeast two-hybrid screen, two groups independently identified a cell wall pectin methylesterase (PME) based on its interaction with the TMV 30-kDa protein (9, 11). Yeast two-hybrid screens have also been used to identify a variety of host proteins that interact with a range of viral movement proteins. These have included DnaJ family members that interact with Tomato spotted wilt virus NSm (48); an apparent plant homolog of the rat vesicular trafficking protein PRA1 reported to interact with the Cauliflower mosaic virus (CaMV) gene I-encoded protein (21); a protein containing RGD motifs characteristically found in several extracellular animal proteins that interact with several members of the integrin family of cell surface matrix receptors, which was reported to interact with the Turnip crinkle virus p8 protein (29); a homeodomain-containing protein reported to interact with Tomato bushy stunt virus P22 (10); three ankyrin-repeat-containing proteins that can interact with the smallest (12K) of the triple gene block movement proteins encoded by Potato virus X (12); a receptor-like protein kinase that binds to Tomato golden mosaic virus and Tomato crinkle leaf yellows virus NSP (30); and a plant-specific acetyltransferase, AtNSI, that interacts with CLCV and SqLCV NSP (32). However, only two of these studies demonstrated the relevance of the reported interaction to virus movement and infectivity in vivo. TMVs expressing deleted forms of the 30-kDa protein impaired in the ability to bind the identified PME are defective in cell-to-cell movement and are not infectious, and TMV systemic spread is delayed in tobacco plants when PME expression is suppressed (8, 9). CLCV infectivity is enhanced on transgenic 35S::AtNSI Arabidopsis lines that overexpress AtNSI (32).
The nuclear trafficking of NSP requires two classic basic nuclear localization sequences (NLS-A and NLS-B) (46) and a leucine-rich nuclear export sequence (NES) of the type found in rapidly shuttling nuclear proteins such as human immunodeficiency virus Rev and Xenopus laevis TFIIIa (Fig. 1) (56). NSP function is also highly regulated by posttranslational modifications and through its interactions with other viral proteins, which include the viral coat protein (CP) as well as MP (24, 39, 41). SqLCV CP can bind, without encapsidating, replicated progeny ssDNA genomes to sequester them away from the nuclear replication pool and make them available for NSP-mediated export from the nucleus (41). It is this synergy between CP and NSP that AtNSI, through its interaction with NSP, appears to regulate.
|
To identify the region of NSP that interacts with AtNSI and to establish that this interaction is required for virus infection, we used a reverse yeast two-hybrid selection and deletion analysis to identify CLCV NSP mutants that were defective in AtNSI binding. We report here that the 38-amino-acid (aa) region between CLCV NSP residues 150 and 187 is necessary for AtNSI binding. Three NSP missense mutants that were impaired in their ability to bind AtNSI were analyzed in detail and found to be comparable to wild-type NSP in their stability in plant cells, nucleocytoplasmic shuttling, DNA binding properties, and ability to interact with CLCV MP. Nevertheless, CLCV that was engineered to express each of these mutated forms of NSP showed reduced levels of infection on Arabidopsis, accumulated lower levels of viral DNA in systemic leaves, and produced delayed and attenuated symptoms, and the extent of these defects generally correlated with the measured defects in NSP binding to AtNSI. Our studies establish the important role of NSP-AtNSI interaction and AtNSI activity in bipartite geminivirus infection and pathogenicity.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
200 leu2-3,112 trp1-901 LYS2::GAL1-HIS3 ade2::GAL2-ADE2 met::GAL7-lacZ) and then plated on dropout medium (15). Histidine prototrophs were recovered and further analyzed in medium lacking adenine or histidine and supplemented with 5 mM 3-aminotriazole. Analyses of ß-galactosidase (ß-Gal) activity were performed by use of a filter lift assay (34).
Single amino acid changes that disrupted the interaction of NSP and AtNSI were identified in a reverse two-hybrid screen using yeast strain MaV203 (MAT
leu2-3,112 trp1-901 his3200 ade2-101 cyh2r can1r gal4 gal80 GAL1::lacZ HIS3UASGAL1::HIS3@LYS2 SPAL10UASGAL1::URA3; Invitrogen) as previously described (6, 51, 53). Mutagenic PCR was performed without MnCl2, using pBI880-NSP as a template and with primers annealing 
200 bp upstream and downstream of the NSP coding region to allow for homologous recombination. Transformants were plated on synthetic complete medium lacking leucine and tryptophan (SC-L-T) containing 0.1% 5-fluoroorotic acid (5-FOA), and a library was generated and further characterized. Selected clones were grown on SC-L-T, and DNAs were extracted (2) and retransformed into yeast strains MaV203 and PJ69-4A. The clones were analyzed by plating on SC-L-T with 0.1% 5-FOA (MaV203) or on SC-H-L-T with or without 5 mM 3-aminotriazole (PJ69-4A). Liquid ß-Gal assays were performed with strain PJ69-4A as reported by Montano (34). Cultures grown to saturation in dropout medium were serially diluted 10-fold in sterile water, and 5 µl of each dilution was assayed.
GST pull-down and ssDNA binding assays. Wild-type or mutated NSP, each cloned as a transcriptional fusion to the T7 promoter in pGEM7Z(+), were labeled with [35S]methionine by coupled in vitro transcription and translation, using the TnT reticulocyte lysate system according to the manufacturer's instructions (Promega). For in vitro protein binding, 5 µl of the appropriate [35S]Met-labeled NSP protein was diluted with a solution containing 20 mM HEPES (pH 7.4), 150 mM KCl, 5 mM MgCl2, 0.1% Triton X-100, 0.1% Nonidet P-40, and 0.1% gelatin (35) to a final volume of 150 µl and then incubated with 2 µg of purified glutathione S-transferase (GST) or GST-AtNSI tethered to glutathione-Sepharose for 2 h at 4°C with gentle agitation (32). After four 10-min washes with 1 ml of phosphate-buffered saline-dithiothreitol (35), the samples were denatured in sodium dodecyl sulfate and resolved by electrophoresis on 12% polyacrylamide gels (SDS-12% PAGE) (45) and detected by autoradiography using En3Hance (Perkin-Elmer).
In vitro binding reactions of NSP and its mutants to ssDNA-cellulose (USB) were performed as previously described (39).
Transient expression studies in protoplasts and leaves. To determine the subcellular localization of the NSP missense mutants, we generated N-terminal green fluorescent protein (GFP) fusions by PCR to translationally fuse the coding sequence for each mutant to that for GFP in the vector pCKGFPS65C, which drives transcription from the enhanced 35S promoter (42). Each fusion was constructed with an NH2-Gly-Gly-Gly-Gly-Ser-COOH linker between NSP-HIS6 and GFP. p35S::MPB was created by using NcoI and BamHI to insert the coding region for MP in place of that for GFP in pCKGFPS65C. For NSP accumulation studies, the NSP-HIS6 coding sequence was cloned in place of that for GFP in pCKGFPS65C by the use of NcoI and XbaI to create p35S::NSP-HIS6.
Arabidopsis leaf protoplasts were prepared and transfected by using polyethylene glycol (PEG) as described previously (1). For preparations of Nicotiana benthamiana protoplasts, leaf strips (2 mm long) were floated on 8 mM CaCl2 in 0.66 M mannitol for 30 min and digested overnight by gentle shaking (30 rpm on a platform shaker) in 0.2% cellulase and 0.05% macerozyme in the dark. Cells were then released from undigested tissue by increasing the speed to 80 rpm for 10 min and were collected by low-speed centrifugation in an IEC clinical centrifuge (model CL) (2 min, 210 x g). The protoplasts were resuspended in 8 mM CaCl2 in 0.66 M mannitol, layered over a cushion of 21% sucrose in water, and centrifuged for 8 min at 440 x g, after which the intact cells were collected from the interface (S. M. Gray, personal communication). After a wash in 8 mM CaCl2 in 0.66 M mannitol, the protoplasts were washed, for 5 min each, in 0.53, 0.46, and finally 0.4 M mannitol and then transfected by using PEG (1). All of these and subsequent centrifugation steps were performed for 2 min in an IEC CL centrifuge at 70 x g. For visualization of GFP fusion proteins, live protoplasts were placed on glass slides under coverslips and were imaged with a Leica TCS SP2 confocal laser scanning microscope.
For agroinfiltration of N. benthamiana leaves, the complete NSP expression cassette was excised from each p35S:NSP-HIS6 construct by digestion with HindIII (see above) and then was cloned into pCB301 (59). These clones were transformed into Agrobacterium tumefaciens strain GV3101 (25) by electroporation. The cultures were grown to an optical density of 0.5, and agroinfiltration was performed as previously described (54). Samples were collected after 72 h and processed as described by Weigel and Glazebrook (58). To determine the levels of NSP accumulation, protoplasts were collected after 24 h and resuspended in Tris-EDTA containing 50 mM NaCl. After the addition of SDS-containing loading buffer and heating for 5 min at 100°C, the samples were resolved by SDS-PAGE on 12.5% gels, blotted onto nitrocellulose, and detected with an anti-His6 monoclonal antibody (Invitrogen) by chemiluminescence (Amersham Biosciences) (38, 45).
Infectivity assays. The CLCV A and B components, including two common regions, were PCR amplified from pGEM7Zf(+) vectors (20), and each of these was cloned separately into the XbaI and KpnI sites of the binary vector pCB301 (59). For the introduction of each NSP missense mutation, the 546-bp AclI-HpaI fragment from NSPE150G, NSPI164T, or NSPD187G was cloned in place of the corresponding fragment in wild-type NSP. The resulting constructs were transformed into A. tumefaciens strain GV3101 (25) by electroporation and were used to inoculate Arabidopsis ecotype Col-0 12-day-old seedlings. For the inoculation procedure, the first two true leaves were cut transversally with a razor blade and dipped for 5 min into a suspension of GV3101, which were overnight cultures concentrated 50x and mixed to contain the appropriate CLCV A and B component clones. Plants were grown at 22°C under a 16 h-8 h light-dark cycle. DNAs were extracted from systemically infected leaves by use of a DNeasy plant kit (Qiagen) as directed by the manufacturer and were quantified by semiquantitative PCR using NSP- or AtNSI-specific (internal control) primers (58).
DNA sequence analysis. DNA sequence analysis was performed by the Cornell University BioResource Center on an Applied Biosystems Automated 3730 DNA analyzer.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
To initially identify the regions of NSP that interact with AtNSI, we tested N- and C-terminal truncations of CLCV NSP for their ability to interact with AtNSI in a classic yeast two-hybrid interactive screen or an in vitro GST pull-down assay. Both assays gave the same results. In the yeast two-hybrid screen, both full-length NSP and NSP lacking the first 42 aa (NSP43-256), when fused to the GAL4 DNA binding domain, interacted with AtNSI fused to the GAD (Fig. 2A). None of the other NSP truncations interacted with AtNSI in this assay, indicating that the N-terminal 42 aa of NSP were dispensable for interacting with AtNSI. The same results were obtained using the in vitro GST pull-down assay, in which full-length NSP and NSP43-256 were the only forms that specifically interacted with GST-AtNSI tethered to glutathione-Sepharose resin (Fig. 2A).
|
We randomly mutagenized NSP fused to the GAL4 DNA binding domain in pBI880 by PCR, using Taq polymerase in the absence of MnCl2 and primers that annealed 200 nucleotides upstream and downstream of the NSP insert. Gap repair transformation with linearized pBI880 was then used to introduce the PCR products into yeast strain MaV203, which contained the AtNSI coding region fused to the GAD in plasmid pBI771 (pBI771-AtNSI) (6), and mutants were selected by growth on dropout medium containing 0.1% 5-FOA. To confirm the results, we transformed plasmid DNAs isolated from surviving colonies into MaV203 harboring pBI771-AtNSI and retested them for growth on 0.1% 5-FOA. The isolated DNAs were also transformed into yeast strain PJ69-4A harboring pBI771-AtNSI to quantify the degree to which the NSP-AtNSI interaction was impaired based on the levels of ß-Gal activity, and each NSP insert was sequenced to identify the mutation(s).
We initially screened for NSP mutants that had completely lost the ability to interact with AtNSI, based on the most vigorous growth of MaV203 on 5-FOA in the reverse yeast two-hybrid screen and only background levels of ß-Gal activity in the classic interactive yeast two-hybrid screen with strain PJ69-4A. These screens generated only truncations or deletions of NSP. We therefore concentrated on characterizing NSP mutants that were defective, but not fully null, in their interaction with AtNSI, as evidenced by their decreased but detectable levels of ß-Gal activity. We recovered three NSP missense mutants that were impaired in their ability to interact with AtNSI, specifically NSPE150G, NSPD187G, and the double mutant NSPF29S/I164T (Fig. 1 and 2). Subcloning of each mutation in NSPF29S/I164T showed that NSPI164T was responsible for the impaired interaction with AtNSI (Fig. 2). Based on their ß-Gal activities as well as the relative levels of growth of PJ69-4A on selective media, NSPE150G was the most defective mutant in its ability to interact with AtNSI, followed by NSPI164T and NSPD187G, with the last being the least defective in its interaction with AtNSI (Fig. 2B).
These findings were confirmed biochemically with an in vitro GST pull-down assay in which [35S]Met-labeled in vitro-transcribed and -translated wild-type NSP, or each NSP missense mutant, was incubated with GST-AtNSI or GST tethered to glutathione-Sepharose resin (Fig. 2C). In this more direct in vitro binding assay, in which equal amounts of NSPE150G, NSPI164T, and NSPD187G were directly compared with wild-type NSP for their ability to bind to a standardized amount of AtNSI, all three missense mutants exhibited notably reduced levels of binding to AtNSI. These in vitro binding results correlated qualitatively with those of the yeast two-hybrid screen and ß-Gal assays in showing that NSPE150G, NSPI164T, and NSPD187G were each defective in binding AtNSI. These findings, combined with the analysis of NSP truncations, showed that the 38-residue region of NSP from aa 150 to 187 was necessary for binding to AtNSI. The mutations in NSPE150G and NSPI164T lie within a region of the protein to which no function had so far been assigned (residues 120 to 183) (46, 56). The mutation in NSPD187G is located within the NES (aa 184 to 193), although it affects a residue that has not been shown to be essential for NES function (see below) (13, 56).
CLCV expressing NSPE150G, NSPI164T, or NSPD187G exhibits lower levels of infectivity and attenuated symptoms on Arabidopsis. To demonstrate that the NSP-AtNSI interaction is necessary for virus infection in vivo, we introduced each of the missense mutations in NSPE150G, NSPI164T, and NSPD187G into the CLCV B component, and the resulting mutated viral DNAs were coinoculated with the wild-type CLCV A component onto 12-day-old Arabidopsis seedlings (Col-0) by agroinoculation (20). Each of the NSP mutants showed lower levels of infection on Arabidopsis (Table 1; Fig. 3A) than wild-type CLCV, which had an average infectivity level of 97%, as assessed by the appearance of disease symptoms. CLCV-NSPI164T always gave somewhat higher levels of infection (85% on average) than CLCV-NSPE150G and CLCV-NSPD187G, which had average infectivity levels of 78 and 75%, respectively.
|
|
Correlating with their relative defects in infectivity levels, each of the NSP mutants also produced attenuated disease symptoms (Fig. 4). For each mutant, the degree of attenuation in symptoms was correlated with the relative decrease in systemic levels of viral DNA (Fig. 3B). CLCV-NSPI164T, which was the most infectious of the mutants based on both its efficiency of infection and its systemic accumulation of viral DNA, produced the most severe symptoms, although these were clearly milder than those produced by wild-type CLCV. CLCV-NSPD187G produced the mildest symptoms, with the symptoms produced by CLCV-NSPE150G being intermediate in severity between those produced by CLCV-NSPI164T and CLCV-NSPD187G (Fig. 4). Fitting with their striking attenuation of disease symptoms and lower infectivity levels, CLCV-NSPE150G and CLCV-NSPD187G also exhibited delays of 1 to 2 days in the initial appearance of disease symptoms (Fig. 3A).
|
The relative levels of NSP accumulation were determined by transient expression assays in N. benthamiana-transfected protoplasts or agroinfiltrated leaves in which the wild-type or mutated forms of NSP were each expressed as a transcriptional fusion to the CaMV 35S promoter and detected by Western blot analyses of cell extracts. Since our available NSP antisera were raised against SqLCV NSP and weakly cross-reacted with CLCV NSP, we cloned each NSP construct as a translational fusion to a C-terminal His6 tag and used an anti-His6 monoclonal antibody for these studies. We did not use GFP-NSP fusions for these analyses as the GFP moiety could have stabilized the mutated forms of NSP. In both of these transient assays, the levels to which CLCV NSPE150G-HIS6, NSPI164T-HIS6, and NSPD187G-HIS6 accumulated were comparable to the levels of accumulation of wild-type CLCV NSP-HIS6, and there was no direct correlation between these relative levels of accumulation and the infectivity levels for each NSP mutant (Fig. 5A and data not shown).
|
90% of the input). A minor amount of wild-type or each mutated NSP eluted in 400 mM KCl, with the vast majority of the bound protein eluting in 1 M KCl (Fig. 5B).
To assess the ability of NSPE150G, NSPI164T, and NSPD187G to target the nucleus, we fused the N terminus of each missense mutant and wild-type NSP in frame to GFP and transiently expressed each protein from the CaMV 35S promoter in transfected N. benthamiana or Arabidopsis (Col-0) protoplasts. Examination by confocal laser scanning microscopy (CLSM) showed that in contrast to free GFP, which was found throughout the cells, CLCV GFP-NSPE150G, GFP-NSPI164T, and GFP-NSPD187G all accumulated in the nuclei of both cell types, and based on the total number of fluorescing cells, they did so to the same extent as wild-type CLCV GFP-NSP (Fig. 6A and data not shown). No differences in the timing or apparent level of nuclear accumulation of each NSP missense mutant compared to wild-type NSP were detected. Fifty to sixty percent of transfected N. benthamiana protoplasts and 15 to 20% of transfected Arabidopsis protoplasts showed nuclear accumulation of the wild type or each mutated GFP-NSP fusion, with nuclear fluorescence first appearing at 14 to 16 h posttransfection and peaking at 24 h posttransfection. Both N- and C-terminal NSP fusions to GFP behaved the same in this assay, completely accumulating in the nucleus.
|
To show that NSPE150G, NSPI164T, and NSPD187G were not defective in either nuclear export or their interaction with MP, we used the ability of MP, when it is coexpressed with NSP in protoplasts, to trap NSP in the cytoplasm and redirect it to the cortical ER to examine both steps simultaneously (46, 47). Previous studies of SqLCV have shown that NSP mutants that are defective in nuclear export or in the ability to interact with MP will remain in the nucleus when coexpressed with wild-type MP (46, 56). N. benthamiana protoplasts were cotransfected with plasmids that expressed CLCV MP and either wild-type CLCV GFP-NSP, or CLCV GFP-NSPE150G, GFP-NSPI164T, or GFP-NSPD187G, and were examined at different times posttransfection for the accumulation of mutated or wild-type GFP-NSP at the cortical ER due to the presence of MP. Protoplasts singly transfected with each GFP-NSP construct served as controls. As early as 16 to 24 h posttransfection, the wild type and each of the mutated CLCV GFP-NSP proteins were seen to be redirected from the nucleus to the cortical ER in the presence of CLCV MP (Fig. 6A). No differences were observed in the timing of this interaction or in the relative levels to which wild-type CLCV GFP-NSP, or GFP-NSPE150G, GFP-NSPI164T, or GFP-NSPD187G accumulated at the cortical ER when coexpressed with MP (Fig. 6A and data not shown). In earlier SqLCV studies, the interaction of NSP with MP in the cortical region of transfected cells was detected by indirect immunofluorescence staining of fixed protoplasts, in which the ER structure was not well preserved (46, 56). The use of GFP-NSP fusions to monitor the interactions allowed us to clearly demonstrate the interaction of CLCV GFP-NSP with CLCV MP at the cortical ER by examining a superimposed confocal Z series taken of live protoplasts (Fig. 6B).
Thus, CLCV NSPE150G, NSPI164T, and NSPD187G are not defective in nuclear export, nor are they defective in their ability to interact with CLCV MP. These findings, combined with the facts that NSPE150G, NSPI164T, and NSPD187G behave like wild-type CLCV NSP in their stabilities and nuclear accumulation in plant cells and their in vitro ssDNA binding properties and that each individual missense mutation was responsible for the observed infectivity defects, lead us to conclude that the NSP-AtNSI interaction is necessary for CLCV infection.
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The strength of the reverse yeast two-hybrid selection is its unbiased nature. This allows one to define the binding region within a protein that is involved in a specific interaction (51, 52). Using this system, we identified three missense mutations in NSP—-NSPE150G, NSPI164T, and NSPD187G—-that impaired its ability to interact with AtNSI both in the yeast two-hybrid screen and in vitro, as determined by a GST pull-down assay (Fig. 2B and C). These mutations implicate the region of NSP that spans residues 150 to 187, which strikingly had no previously assigned specific function, as containing the AtNSI binding site. Our analysis of N- and C-terminal truncations of NSP confirmed that the N-terminal 42 residues, which contained the bipartite NLS and part of the ssDNA binding region, were not required for AtNSI binding (Fig. 1 and 2). However, beyond this none of the other NSP truncations could interact with AtNSI. The most likely explanation for this failure is that the majority of deletions caused global misfolding of NSP. In support of this, our previous analyses of SqLCV NSP showed that a C-terminal deletion (195-256) impaired the nuclear targeting of NSP, despite the fact that the NSP NLSs are located within the NSP N terminus at residues 21 to 42 and 81 to 96 (46).
Correlated with their impaired ability to bind AtNSI, the three NSP missense mutants NSPE150G, NSPI164T, and NSPD187G clearly decreased the pathogenicity of CLCV for Arabidopsis: viral disease symptoms were delayed and notably attenuated for these mutants, and the efficiency of virus infection on this host was correspondingly reduced (Fig. 3 and 4; Table 1). Both of these features are consistent with virus movement being impaired, and they have been characteristically observed in analyses of SqLCV MP or NSP mutants (23, 24, 46, 47). The decreased efficiency of virus infection was confirmed for each NSP mutant by quantifying the relative levels of viral DNA in extracts of systemically infected leaves (Fig. 3). The observed decreases in the levels of systemic DNA for each NSP mutant correlated with the degrees to which CLCV symptoms were attenuated and infectivity was decreased, a finding that again is consistent with a defect in virus movement. In addition, the extents to which CLCV symptom production and infectivity levels were affected by NSPE150G, NSPI164T, and NSPD187G were similar both in the degree of the effects on disease appearance and the efficiencies of infection observed on natural hosts when bipartite geminivirus CP was eliminated (3, 5, 14, 18, 23, 24, 40) and in the magnitude of change observed in the efficiency of CLCV infection on Arabidopsis lines that overexpress AtNSI (32). Eliminating virus CP does not abolish infectivity, although it does abolish whitefly transmission: depending on the particular bipartite geminivirus and plant host combination, the effects can vary from a simple delay of 1 to 2 days in the appearance of disease symptoms to a delay in symptom appearance of several days accompanied by attenuated disease symptoms and a decrease in the efficiency of infection by 40 to 50% (3, 5, 14, 18, 23, 24, 40). The fact that wild-type CLCV infectivity is enhanced 30% on Arabidopsis lines that overexpress AtNSI from a 35S::AtNSI transgene is directly relevant here (32). Our findings that CLCV mutants that expressed NSPE150G, NSPI164T, and NSPD187G produced delayed and attenuated symptoms, and were 15 to 30% less efficient than wild-type CLCV in infecting Arabidopsis, are consistent with these results. Thus, rather than pathogenicity being an all-or-nothing situation, it appears that bipartite geminiviruses may utilize host metabolic or regulatory pathways in a variety of ways to maximize their ability to infect the host, a conclusion that was also suggested by the recent demonstration that the transcriptional transactivator protein TrAP encoded by tomato golden mosaic virus interacts with and inhibits the host kinases SNF1 and ADK (17, 55).
While mutational studies have shown that the yeast two-hybrid screen and associated ß-Gal assay can be used to examine defects in a known protein-protein interaction, the results in general correlate qualitatively, but not necessarily quantitatively, with those obtained when the same interaction is studied in vivo in the relevant cell type (for examples, see references 34 and 49). In our examination of NSPE150G, NSPI164T, and NSPD187G, we found that the degree to which each mutant was defective for infectivity in planta correlated qualitatively, but not always quantitatively, with its defect in interacting with AtNSI, as determined with the yeast two-hybrid assay. NSPE150G was the most defective mutant in interacting with AtNSI in this assay and NSPI164T was less defective, and this did directly correlate with NSPE150G being less infectious, accumulating less viral DNA in systemically infected leaves, and producing more attenuated symptoms on Arabidopsis than did NSPI164T (Fig. 2 to 4). However, although NSPD187G appeared to have the least defective AtNSI interaction, as determined by the yeast two-hybrid and ß-Gal assays, it was as defective as NSPE150G in its infectivity on Arabidopsis and its systemic accumulation of viral DNA, and it produced somewhat more attenuated symptoms.
Because our heterologous SqLCV NSP antisera reacted weakly with CLCV NSP, we could not readily detect the accumulated levels of NSPE150G, NSPI164T, and NSPD187G in immunoblots of yeast cell extracts. Thus, it is possible that some of these differences were due to different levels of accumulation of each NSP mutant in the yeast two-hybrid assay. The in vitro GST pull-down assay allowed us to examine these defects in binding to AtNSI directly and more quantitatively. In this direct in vitro binding assay, NSPE150G, NSPI164T, and NSPD187G were all similarly defective in their ability to bind AtNSI (Fig. 2C). A truly quantitative in vitro binding assay, such as a filter binding assay, would more precisely examine the relative defects in AtNSI binding for NSPE150G, NSPI164T, and NSPD187G.
Given the large number of interactions in which NSP is involved, it was critical to establish that NSPE150G, NSPI164T, and NSPD187G were specifically defective only in their interaction with AtNSI. Clearly, a missense mutation could have affected the stability of NSP, its interactions with the nuclear import or export machinery of the cell, its cooperative interaction with MP, or its binding to viral genomes, any one of which would affect virus movement and the efficiency of infection. NSPE150G, NSPI164T, and NSPD187G each accumulated to the same level as wild-type NSP when expressed in N. benthamiana leaves or protoplasts (Fig. 5A and data not shown), which indicated that none of the mutations had a significant effect on the stability of NSP in plant cells. Furthermore, NSPE150G, NSPI164T, and NSPD187G were indistinguishable from wild-type NSP in their in vitro ssDNA binding properties (Fig. 5B); thus, defects in DNA binding did not account for the infectivity defects of these missense mutants.
To examine nuclear targeting, we used CLSM to visualize the nuclear accumulation of GFP-NSP fusions in Arabidopsis and N. benthamiana protoplasts. We used these same constructs, coexpressed with MP in protoplasts, to examine both nuclear export and the interaction with MP based on MP trapping NSP in the cytoplasm, an assay that was used to identify the NES in SqLCV NSP (56). An additional advantage of using the latter assay in living protoplasts was that it allowed us to both visualize the interaction with MP based on the fluorescence of the GFP-NSP fusions and clearly image this interaction at the cortical ER (Fig. 6).
Based on the timing of their appearance and the levels of GFP fluorescence observed, we detected no differences in the nuclear accumulation of NSPE150G, NSPI164T and NSPD187G compared to wild-type NSP, nor did we observe differences from wild-type NSP in their nuclear export and interaction with MP based on our coexpression assay (Fig. 6). This last result clearly showed that the mutation in NSPD187G, although it was within the predicted NES in CLCV NSP, was not essential for NES function. These studies further establish that CLCV NSP and MP behave in the same manner as do SqLCV NSP and MP in their subcellular localization to nuclei or the cortical ER, respectively, and their cooperative interaction to trap NSP in the cytoplasm and direct it to the cortical ER (Fig. 6).
In contrast to results from previous studies in which SqLCV NSP was completely redirected to the cortical cytoplasm when it was coexpressed with SqLCV MP in Nicotiana tabacum cv. Xanthi protoplasts (46, 47), similar residual levels of wild-type and mutated nuclear GFP-NSP remained at 24 h in the presence of MP in N. benthamiana protoplasts (Fig. 6). The earlier studies used indirect immunofluorescent staining of fixed protoplasts to localize SqLCV NSP as well as MP, and sufficiently high levels of NSP or MP were first detected at 48 h posttransfection. The transfection rates for the electroporated N. tabacum cv. Xanthi protoplasts were also lower, being 10 to 15% compared to the 50 to 60% we obtained here for N. benthamiana protoplasts transfected by the use of PEG. Combined with the increased sensitivity afforded by GFP, the higher transfection efficiency of N. benthamiana protoplasts may have led to higher levels of accumulation of our GFP-NSP fusion proteins, which allowed us to readily detect these fusions between 16 and 24 h posttransfection in these living cells. However, it may also have saturated the nuclear export machinery.
Consistent with NSP and MP accumulating more rapidly and to higher levels in the transfected N. benthamiana cells, it was difficult to image these cells after 36 h, as they rapidly began to lyse probably due to the documented toxicity of MP (39, 47). However, we cannot exclude the possibility that GFP had some effect on the efficiency with which the GFP-NSP fusions were exported from the nucleus. If there was some effect, it was specific to GFP since the nuclear export and interaction with MP for all three NSP missense mutants were indistinguishable from those for wild-type NSP. It is also possible that some, if not all, of the residual nuclear fluorescence was accumulated free GFP cleaved from the fusion proteins, which has been reported in other studies (for example, see reference 4).
Based on these studies, we conclude that NSPE150G, NSPI164T, and NSPD187G are specifically defective in their binding to AtNSI. Consistent with this, none of these three mutations had previously been identified in our extensive analyses of SqLCV NSP in host range or its DNA binding activity, nuclear trafficking, and interaction with MP (22-24, 46, 56). Thus, the effects of NSPE150G, NSPI164T, and NSPD187G in attenuating disease symptoms and decreasing the infectivity of CLCV on Arabidopsis appear to be the consequence of their specific defect in interacting with AtNSI. This establishes a role for NSP-AtNSI interaction in CLCV pathogenesis and the virus's ability to move in and infect Arabidopsis cells. Our earlier studies showed that AtNSI also interacts with SqLCV NSP, which suggested a more general role for this interaction in bipartite geminivirus infection. In this context, it is interesting that NSPE150G, NSPI164T, and NSPD187G identify three residues that are highly conserved among the NSP proteins encoded by a wide range of bipartite geminiviruses (Fig. 1). From the sequences analyzed, only Ile164 is seldom changed to Val, which is also a neutral hydrophobic amino acid; Glu150 and Asp187 are completely invariant. This further supports the conclusion that NSP-AtNSI binding plays an important role in the bipartite geminivirus life cycle.
The findings reported here support and extend our earlier studies on NSP-AtNSI interaction and its role in virus infection. Based on our previous results, CP can synergistically interact with NSP by binding replicated viral ssDNA, without encapsidating it, to sequester the viral genome away from the replication pool and thus make it available for nuclear export (41). We have proposed that NSP, bound to the viral ssDNA genome, recruits AtNSI to a ternary complex to acetylate the genome-bound CP. Analogous to the proposed role of histone acetylation in chromatin remodeling during transcriptional activation (7), this would decrease CP avidity for the viral ssDNA and thus allow NSP to displace CP as NSP cooperatively binds the genome for export from the nucleus (32). Several observations are consistent with this model. AtNSI acetylates CLCV CP in vitro, yet AtNSI and CP do not stably interact themselves. In addition, the overexpression of AtNSI from a 35S::AtNSI transgene enhances the level of CLCV infection of Arabidopsis (32). This would be expected since CP itself does not stably interact with AtNSI; thus, high levels of AtNSI should not affect CP first binding to replicated genomes. However, the high levels of AtNSI would be expected to increase the rate at which it is recruited by NSP to acetylate CP, thereby increasing the rates of CP release and nuclear export of the viral genome. Our model further predicts that NSP mutants such as NSPE150G, NSPI164T, and NSPD187G, which are impaired in their ability to bind AtNSI, would be defective in recruiting AtNSI to displace CP from the viral genome and thus would be less efficient in exporting the viral ssDNA from the nucleus. This defect in movement would be evident as attenuated symptoms and decreased levels of infectivity, as we observed for CLCV-NSPE150G, CLCV-NSPI164T, and CLCV-NSPD187G.
|
ACKNOWLEDGMENTS |
|---|
This work was supported by National Science Foundation grant MCB-9982622 to S.G.L. and a PRAXIS XXI (FCT, MCES, Portugal) predoctoral fellowship to M.F.C.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
