Interactions of the TGB1 Protein during Cell-to-Cell Movement of Barley Stripe Mosaic Virus

Abstract

We have recently used a green fluorescent protein (GFP) fusion to the γb protein of Barley stripe mosaic virus (BSMV) to monitor cell-to-cell and systemic virus movement. The γb protein is involved in expression of the triple gene block (TGB) proteins encoded by RNAβ but is not essential for cell-to-cell movement. The GFP fusion appears not to compromise replication or movement substantially, and mutagenesis experiments demonstrated that the three most abundant TGB-encoded proteins, βb (TGB1), βc (TGB3), and βd (TGB2), are each required for cell-to-cell movement (D. M. Lawrence and A. O. Jackson, Mol. Plant Pathol. 2:65–75, 2001). We have now extended these analyses by engineering a fusion of GFP to TGB1 to examine the expression and interactions of this protein during infection. BSMV derivatives containing the TGB1 fusion were able to move from cell to cell and establish local lesions in Chenopodium amaranticolor and systemic infections of Nicotiana benthamiana and barley. In these hosts, the GFP-TGB1 fusion protein exhibited a temporal pattern of expression along the advancing edge of the infection front. Microscopic examination of the subcellular localization of the GFP-TGB1 protein indicated an association with the endoplasmic reticulum and with plasmodesmata. The subcellular localization of the TGB1 protein was altered in infections in which site-specific mutations were introduced into the six conserved regions of the helicase domain and in mutants unable to express the TGB2 and/or TGB3 proteins. These results are compatible with a model suggesting that movement requires associations of the TGB1 protein with cytoplasmic membranes that are facilitated by the TGB2 and TGB3 proteins.

The requirements for systemic plant virus infections differ from those of animal viruses because the plant cell wall poses a barrier that must be breached to permit initial infections, cell-to-cell movement, and entry into and exit from the vasculature. Localized spread from primary infection foci requires that plant viruses move to adjacent cells through the intercellular channels formed by plasmodesmata (for reviews, see references 11, 35, and 37). Consequently, most plant viruses have evolved dedicated genes that function to mediate cell-to-cell transit. A considerable body of evidence now exists indicating that many of these viruses encode genes that function mechanistically to increase the permeability of the plasmodesmata and to facilitate transport of bound viral nucleoprotein complexes through the plasmodesmata to adjacent cells (1, 17, 30, 31, 41, 61).

Tobacco mosaic virus (TMV) is the prototype of those viruses that encode a single cell-to-cell movement protein, and the movement processes described for TMV have provided valuable models for analysis of the infection processes of other viruses (28, 49). In many viruses, a single movement protein facilitates cell-to-cell transit by a series of incompletely defined steps during which the protein encapsidates the viral genome and associates with cytoplasmic membranes, cytoskeletal elements, and cell wall proteins. These events alter plasmodesmal permeability sufficiently to permit cell-to-cell transport of putative viral nucleoprotein complexes (40). However, numerous other viruses require coordinated activities of more than one gene for local and long-distance transport. For example, in the case of the DNA-containing geminiviruses, a dual shuttle gene system mediates transit of single-stranded (ss) DNA from the nucleus, through the cytoplasm, and into and across the plasmodesmata (28, 41). Two small proteins, p8 and p9, are required for systemic infections of the cytoplasmically replicating carmoviruses (29), and the tombusviruses have two nested genes (p19 and p22) that function in different aspects of movement (55). The Tomato bushy stunt virus p22 protein appears to be the functional analog of the TMV 30-kDa movement protein, whereas the p19 protein is dispensable for systemic movement in some host backgrounds but is required in other hosts (55). Additional complexity and division of labor is observed among a diverse set of RNA viruses differing substantially in genome organization that encode movement proteins organized in a “triple gene block” (TGB). These proteins have homologs in monopartite potexviruses (30, 39) and carlaviruses (51), bipartite benyviruses (18) and pecluviruses (22), and tripartite hordeiviruses (38, 56).

Barley stripe mosaic virus (BSMV) is the type member of the hordeiviruses. The viral genome is composed of positive-sense ssRNA divided into three components designated α, β, and γ. The α and γ RNAs are strictly required for replication, while RNAβ is required for cell-to-cell movement (27, 44). RNAα encodes a replicase protein, αa, that contains methyl transferase and helicase domains. RNAγ is bicistronic and encodes a second replicase protein, γa, that is characterized by a polymerase (GDD) motif. The second protein, γb, is expressed from a subgenomic (sg) RNA (20). This cysteine-rich protein is not required for replication or cell-to-cell movement per se, but mutations within the protein affect pathogenicity (12, 43).

RNAβ encodes a 5′-proximal coat protein (CP) separated from the TGB by a short intergenic region (19, 23). The TGB1 protein, formerly designated βb, is expressed from the 2.45-kb sgRNAβ1 (66). This 58-kDa TGB1 protein has been purified from BSMV-infected barley and shown to bind ssRNA and double-stranded RNA, to exhibit ATPase activity, and to bind nucleotides in vitro (13). A helicase motif is also a prominent component of TGB1, and mutations of conserved amino acids within the motif abrogate cell-to-cell movement (27); however a variety of assays have failed to detect helicase activity in vitro (13). The remaining TGB proteins are expressed from the low-abundance 960-nucleotide sgRNAβ2 (66). TGB2, a 14-kDa protein (formerly designated βd), and its low-abundance 23-kDa translational read-through protein, TGB2′ (formerly designated βd′), are translated by ribosomes initiating at the first AUG of sgRNAβ2. The 17-kDa TGB3 protein (formerly designated βc) is expressed via leaky scanning through the TGB2 AUG codon of sgRNAβ2 (66). Sequence analyses predict that both the TGB2 and TGB3 proteins are membrane associated, as they contain two hydrophobic membrane-spanning domains separated by a hydrophilic region (38, 56). Subcellular fractionation studies have also shown that TGB2 and its read-through extension, TGB2′, are associated with membrane and cell wall fractions (14, 66). Infectivity results clearly demonstrate that the TGB1, TGB2, and TGB3 proteins are each required for cell-to-cell movement in both monocot and dicot hosts (27). These results are similar to and extend those obtained with the potexviruses White clover mosaic virus (2, 30) and Potato virus X (PVX) (60), the benyvirus Beet necrotic yellow vein virus (BNYVV) (4, 18), and the pecluvirus Peanut clump virus (PCV) (22).

Only limited information is available to describe long-distance movement processes whereby viruses enter, navigate and exit the vascular system, but the mechanisms required for these events appear to be distinct from those required for cell-to-cell movement processes (7, 32). For example, efficient long-distance transport of TMV requires the CP in addition to the 30-kDa protein (40). However, a limited number of viruses encoding different classes of movement protein genes can establish efficient local and systemic infections in the absence of the CP. In the prototypical case of Tobacco rattle virus (6) and other tobraviruses encoding a single movement protein (5, 33), the CP is dispensable for long-distance movement. Other examples where local and long-distance transit do not require the CP have been reported among members of the tombusviruses (9, 54).

Considerable variation in the requirements for the CP for local and long-distance movement also exists among those viruses that require the TGB proteins for movement. In the case of BSMV, the CP is dispensable for cell-to-cell and systemic movement (27, 44). Similarly, the CP is dispensable for cell-to-cell movement of BNYVV (15) and PCV (22) and for systemic movement of the pomovirus Potato mop top virus (PMTV) (34). However, unlike BSMV and PMTV, the CP is required for vascular transport of BNYVV (53) and PCV (22). In contrast, potexviruses have an absolute requirement for the CP for cell-to-cell movement (30, 52). To begin to define the cellular interactions during the movement of BSMV, we have constructed green fluorescent protein (GFP) fusions to the amino terminus of TGB1 and have used these reporter derivatives to assess expression patterns and subcellular localization during cell-to-cell movement. In addition, protoplast experiments have been conducted to evaluate the roles of TGB2 and TGB3 in subcellular targeting of TGB1.

MATERIALS AND METHODS

Recombinant plasmids.

Full-length α, β (β42SpI), and γ (γ42) cDNA clones derived from the BSMV ND18 strain (45) were used in this study. GFP was expressed from the γ cDNA clone as a γb-GFP fusion (27). Several site-directed mutations (25) described below were introduced into β42SpI or B7, a mutant of β42SpI that lacks the AUG of the CP (βa) gene (47)_. Additional cDNA clones containing mutations in the RNAβ TGB were used in various experiments. The β_Δ2.0 cDNA clone contains a 200-amino-acid deletion in TGB1 (47), the βCla cDNA clone alters two amino acids (K₁₁ → N and Y₁₂ → R) in TGB2 (47), and the βc-stop cDNA clone introduces a UAA at codon 72 in TGB3 (27). The β cDNA clones that contained mutations in the helicase domain of βb resulted in the following amino acid changes: K₂₇₅ → R (M1); K₂₇₅ → A (M2); D₈₂₆ and E₈₂₇ → N and Q (M3); G₈₄₈, D₈₄₉, and Q₈₅₂ → A, N, and E (M4); R₃₇₇ → A (M5); Q₄₆₂ and G₄₆₃ → E and A (M6); R₄₉₂ → A (M7); and R₃₇₇ and R₄₉₂ → A and A (M5/M7) (13).

To generate the GFP-TGB1 fusion protein, NheI sites (indicated in boldface) were engineered directly after the start codon of TGB1 in β42SpI by site-directed mutagenesis using the primer βb5′NheI (5′ CTTTAGCCATGGCTAGCACGAAAACTG 3′) to generate β42SpI.NheI. The GFP clone pRSGFP-C1 (Clontech, Palo Alto, Calif.) was modified to incorporate NheI sites at the 5′ and 3′ termini by PCR using the primers GFP5′NheI (5′ GGCGCTAGCAGTAAAGGAGAAGA 3′) and GFP3′NheI (5′ CCGGCTAGCTTTGTATAGTTCATC 3′). The PCR product was digested with NheI prior to ligation into NheI-linearized β42SpI.Nhe. The resulting cDNA clone, β.GFP:TGB1, contained the GFP sequence fused in frame to generate a reporter gene for use in expression and cytological experiments. For some experiments, the TGB1 gene in the CP-deficient variant B7 was replaced with the GFP-TGB1 gene from β.GFP:TGB1 by digestion and fragment substitution into the XbaI and SpeI sites to generate β7.GFP:TGB1. These GFP derivatives produced fusion proteins exhibiting a red-shifted excitation peak with increased signal intensity.

Inoculation and treatment of plants and protoplasts.

RNAα, -β, and -γ transcripts prepared as described previously were used to inoculate barley (Hordeum vulgare), Chenopodium amaranticolor, and Nicotiana benthamiana plants (46) or to transfect protoplasts from the BY-2 Nicotiana tabacum cell suspension culture (62). Transfected protoplasts were incubated with cytochalasin D (25 μg/ml) or subjected to a cold treatment prior to being aldehyde fixed and visualized by epifluorescence microscopy (36). The protoplasts were also stained at 24 h posttransfection by incubation in fresh medium containing 10 μM ER-tracker blue-white DPX (Molecular Probes, Eugene, Oreg.) or 0.1 μM rhodamine B hexyl ester (Molecular Probes) for 30 min, transferred to fresh medium without the marker dyes, and visualized under epifluorescence microscopy.

Protein and RNA analyses.

Infected leaves or protoplasts were harvested, extracted, and separated on sodium dodecyl sulfate (SDS)-polyacrylamide gels prior to Western blot serological analysis (14). RNA was extracted, separated on 1% agarose gels, blotted onto nylon membranes, and hybridized with ³²P-labeled β-specific or γ-specific probes as described previously (66).

GFP fusion protein expression and immunofluorescence.

GFP expression was visualized by excitation at 450 to 490 nm and emission at 520 nm using a Zeiss Axiophot (Thornwood, N.Y.) microscope. Computer images were acquired with an Optronics 450 Color charge-coupled device camera and captured at high resolution by using a Scion CG-7 RGB framegrabber board or photographed with Kodak color film. Figures were assembled using Canvas (Deneba Software, Miami, Fla.) and Adobe Photoshop (Adobe Systems Inc., San Jose, Calif.) software. For immunofluorescence, protoplasts were fixed with aldehyde prior to incubation for 1 h at 4°C with a TGB1 polyclonal mouse antibody (1:500 dilution) (35). Following incubation, the protoplasts were washed with phosphate-buffered saline (10 mM phosphate, pH 7.4, 150 mM NaCl) and incubated with a 1:30 dilution of fluorescein-conjugated goat anti-mouse immunoglobulin G (Calbiochem, San Diego, Calif.) for 30 min at 4°C. Residual nonspecific fluorescence was removed by a second phosphate-buffered saline wash prior to microscopic observation.

RESULTS

BSMV GFP-TGB1 reporter virus moves efficiently in monocot and dicot hosts.

To examine the interactions of BSMV TGB1 during infection of plants and protoplasts, we engineered an amino-terminal fusion of the GFP reporter gene to TGB1 (Fig. 1A). Fusions were constructed in both wild-type RNAβ (β.GFP:TGB1) and in the RNAβ mutant B7 (B7.GFP:TGB1), in which CP expression had been eliminated. The β.GFP:TGB1 and B7.GFP:TGB1 derivatives were designed as alternatives to facilitate comparisons of the movement of BSMV reporters in different genetic backgrounds and to provide a high-resolution marker for subcellular localization of GFP-TGB1 during cell-to-cell movement. When the β.GFP:TGB1 (CP-expressing) and B7.GFP:TGB1 (inactivated CP) RNAs were cotransfected with wild-type RNAα and -γ into BY-2 tobacco protoplasts, RNA blotting revealed that the β.GFP:TGB1 and B7.GFP:TGB1 RNA accumulation was reduced by more than 50% compared to levels of wild-type RNAβ or B7 RNAβ (Fig. 1B). To determine whether these decreases in abundance were due to a general reduction in total viral RNA replication, the RNA blots were stripped and reprobed with a γ-specific probe (Fig. 1B). Some batch-to-batch sample variation was observed between electroporated protoplasts. However, RNA blots hybridized with β- and γ-specific probes revealed that similar levels of variation were observed in comparisons of RNAγ and sgRNAγ with the wild-type β or B7 RNAs and wild-type β or B7 RNAs that contained the GFP-TGB1 sequence (Fig. 1B). These results indicate that the smaller amounts of the β.GFP:TGB1 and B7.GFP:TGB1 RNAs are specifically correlated with reduced RNAβ replication rather than reduced viral replication. However, the GFP fusions were expressed to easily detectable levels in protoplasts with antisera raised to either TGB1 or GFP (Fig. 1C).

FIG. 1.

GFP-TGB1 fusion protein expression in BY-2 protoplasts. (A) Diagram of GFP fused to TGB1 in the β RNA or the B7 RNA in which the CP AUG codon was destroyed (*) showing the relative size of the two sgRNAs. (B) Hybridization patterns of RNA from protoplasts transfected with BSMV RNAs at 22 h p.i. The RNAs were separated on agarose gels and blotted onto nylon membranes prior to being hybridized with a ³²P-labeled β-specific probe or a ³²P-labeled γ-specific probe. Lane 1, mock transfection (M); lane 2, α, β (+cp), and γ RNAs; lane 3, α, B7 (−cp), and γ RNAs; lane 4, α, β.GFP:TGB1 (+cp), and γ RNAs; lane 5, α, B7.GFP:TGB1 (−cp), and γ RNAs. Note that the genomic (g) RNAγ is somewhat obscured by rRNAs, but the sgRNAγ is clearly visible. (C) Proteins from protoplasts transfected with BSMV RNAs at 22 h p.i. Extracts were separated on an SDS–10% (wt/vol) polyacrylamide gel prior to immunoblotting with antiserum raised to TGB1 or GFP. Lane 1, mock inoculation (M); lanes 2 and 4, protoplasts transfected with α, β.GFP:TGB1 (+cp), and γ RNAs; lanes 3 and 5, protoplasts transfected with α, B7.GFP:TGB1 (−cp), and γ RNAs; lanes 1, 2, and 3, protoplasts harvested at 19 h posttransfection; lanes 4 and 5, protoplasts harvested at 26 h posttransfection.

For infectivity comparisons of the GFP-TGB1 fusion and wild-type TGB1, α and γ RNAs plus the β.GFP:TGB1 RNA (CP-expressing background) or wild-type β RNA were inoculated onto C. amaranticolor, barley, or N. benthamiana. Our previous observations of systemic symptom development and lesion formation indicated that the γb-GFP derivative was not noticeably compromised in movement compared to wild-type BSMV (27). However, the GFP-TGB1 derivative exhibited a 12- to 24-h delay over the normal 4 to 7 days required for symptom appearance in barley, C. amaranticolor, and N. benthamiana. In addition, the disease phenotype was attenuated in N. benthamiana (not shown). We believe that these minor effects on symptom development are a consequence of the reduced level of viral replication that was observed in protoplasts (Fig. 1B). Nevertheless, these results suggest that the GFP-TGB1 fusion has relatively minor effects on the initial infection events and provides a useful tool for analysis of a variety of events involved in the early stages of BSMV infection.

CP expression affects the virulence of the TGB1 reporter derivatives.

We previously observed similar timing of infection in comparisons of the wild-type BSMV (RNAβ) and its RNAβ B7 CP⁻ derivative (47). To investigate whether the TGB1 fusions affected the infection phenotype, comparisons were carried out with the RNAβ (CP-expressing) and the B7 (inactivated CP) GFP-TGB1 derivatives. These comparisons revealed that the appearance of local lesions in C. amaranticolor was delayed between 24 and 36 h in the GFP-TGB1 B7 background, and these lesions were approximately 50% smaller at 6 days postinoculation (p.i.) (Fig. 2A and B). In addition, the lesions failed to spread and coalesce, and the infected leaves did not undergo abscission by 14 days p.i. as normally occurs in wild-type infections (43). In barley and N. benthamiana, a 36- to 72-h delay in development of systemic symptoms was also observed, but otherwise the infection phenotype appeared to be typical of GFP-TGB1 in a CP-expressing background and of wild-type virus infections (data not shown). These results suggest that even though the GFP-TGB1 derivatives are able to establish systemic infections in the absence of the CP, subtle interactions requiring the CP promote optimal efficiency of both cell-to-cell and systemic movement.

FIG. 2.

Symptom phenotype and protein expression in infections with β.GFP:TGB1 or B7.GFP:TGB1. (A and B) C. amaranticolor leaves inoculated with α and β.GFP:TGB1 (CP+) or B7.GFP:TGB1 (CP−) and γ RNAs at 6 days p.i. (C) Proteins from infected plants separated on an SDS–10% (wt/vol) polyacrylamide gel prior to immunoblotting with antiserum raised to TGB1 or GFP. Lane 1, plants infected with α, β, and γ RNAs; lanes 2, 4, 6, and 8, plants infected with α, β.GFP:TGB1, and γ RNAs (I); lanes 3, 5, 7, and 9, mock-infected plants (M). Lanes 1, 2, 3, 4, and 5, barley; lanes 6 and 7, N. benthamiana; lanes 8 and 9, C. amaranticolor.

Dicot and monocot hosts differ in the expression patterns of GFP-TGB1.

For infectivity comparisons, the wild-type β RNA or β.GFP:TGB1 RNA (CP-expressing background) and the α and γ RNAs were inoculated onto C. amaranticolor, N. benthamiana, and barley. As was previously described in detail (27), the initial infection events differed markedly in the monocot host and the dicot hosts. The accumulation of the fusion proteins was first examined by analysis of protein extracted from infected leaves. The GFP-TGB1 fusion protein was detected in C. amaranticolor, N. benthamiana, and barley by using antiserum raised against the GFP or TGB1 protein (Fig. 2C).

The localization of TGB1 during the infection of C. amaranticolor leaves was followed for 7 days after inoculation with wild-type α and γ RNAs and β.GFP:TGB1 RNA. In time course observations at 24 h p.i., distinct bright-green fluorescence foci composed of two to five epidermal cells could be identified (Fig. 3A). By 48 h p.i., the foci had increased to encompass 8 to 12 epidermal cells (Fig. 3B), and by 120 h p.i., GFP expression could be clearly observed in several hundred epidermal cells (Fig. 3C). At this time, the first signs of lesion development were detected by bright-field microscopy (data not shown), and a small region of bright-yellow fluorescence corresponding to the developing water-soaked regions appeared in the centers of the lesions (Fig. 3C). Green fluorescence could also be observed in the mesophyll cells below the infected epidermal cells, but the three-dimensional movement kinetics were not carefully monitored. During spread of the infection foci, GFP-TGB1 exhibited a distinct temporal gradient of intense fluorescence associated with the outer ring of the developing foci, with a considerable reduction of intensity towards the centers of the lesions (Fig. 3C). These results contrast markedly with γb-GFP fluorescence (compare Fig. 3C and D), which, as previously described, has a uniform pattern of fluorescence throughout the foci (27).

FIG. 3.

Localization of GFP-TGB1 fusion protein in developing infection foci in leaves. (A) Inoculated leaf of C. amaranticolor at 24 h p.i. Bar = 25 μm. (B) Inoculated leaf of C. amaranticolor at 48 h p.i. Bar = 25 μm. (C) Inoculated leaf of C. amaranticolor at 96 h p.i. Note that the center of the lesion is necrotic and that GFP fluorescence is intense at the periphery of the foci some distance away from the necrotic center. Bar = 50 μm. (D) Inoculated leaf of C. amaranticolor with α, β, and γb-GFP RNAs at 96 h p.i. Bar = 50 μm. (E and F) Epidermal C. amaranticolor cells at the outer edge of the infection ring at 72 h p.i. Bar = 50 μm. Inset, bar = 5 μm. Note the paired punctate foci appressed along the cell walls. (G) Inoculated leaf of N. benthamiana at 5 days p.i. Bar = 50 μm. (H) Epidermal N. benthamiana cells at 6 days p.i. Bar = 50 μm. Inset, bar = 5 μm. (I) Inoculated barley leaf at 72 h p.i. Bar = 100 μm. (J) Systemically infected barley leaf at 96 h p.i. Bar = 200 μm.

Throughout the time course of lesion formation in C. amaranticolor, GFP-TGB1 exhibited a diffuse fluorescence within cells that appeared to be associated with membranes, and intense punctate regions of fluorescence occurred at various intervals along the cell walls (Fig. 3A, B, E, and F). Towards the leading edge of the developing lesions, fluorescence often extended across the walls into adjacent cells to produce paired punctate foci (Fig. 3F). To ensure that the localization patterns of GFP-TGB1 in C. amaranticolor were not artifacts resulting from damaged cells undergoing lesion formation, fluorescence was also examined in the systemic hosts, N. benthamiana and barley. In N. benthamiana, green fluorescence was readily observed in the initially infected cells by 24 h p.i., and this was followed by radial expansion into adjacent cells (Fig. 3G). This movement pattern was noted previously in N. benthamiana infections with the γb-GFP reporter virus (27). However, γb-GFP exhibited uniform fluorescence in the developing foci within 3 to 4 days p.i., but with GFP-TGB1, fluorescence was most intense along the leading edges of the infections (Fig. 3G), as was also the case in C. amaranticolor (Fig. 3C). GFP-TGB1 fluorescence was also intense at punctate foci that frequently appeared to traverse the epidermal cell walls of N. benthamiana (Fig. 3H). By 11 days p.i., GFP-TGB1 fluorescence moved into uninoculated N. benthamiana leaves and exited from the veins (not shown). However, the fluorescence of GFP was erratic and was less intense in leaves developing systemic symptoms than in the inoculated leaves. This indicates that as infection progresses the GFP-TGB1 fusion becomes increasingly less stable as the virus moves into the uninoculated leaves. In this regard, immunoblot analysis of protein extracts from uninoculated N. benthamiana leaves revealed the presence of the full-length GFP-TGB1 fusion protein, but forms with deletions could be detected in different samples at various levels (data not shown).

In barley, the GFP-TGB1 fluorescence patterns were quite different from those observed in the dicot hosts. GFP fluorescence was first detected between 72 and 96 h p.i. in the inoculated barley leaves (Fig. 3I), as was previously observed with the γb-GFP reporter virus (27). By 96 h p.i., fluorescence was also observed as a diffuse signal in the systemically infected leaves (Fig. 3J). However, the punctate foci noted along the cell walls of C. amaranticolor and N. benthamiana were not detected in barley, despite extensive searches for their presence (Fig. 3I and J). These host-specific patterns occurring between dicot and monocot hosts complement and extend those previously noted with γb-GFP (27).

Because the experiments described earlier suggest that the CP might affect the kinetics of cell-to-cell movement, we compared the localization of GFP-TGB1 in C. amaranticolor, N. benthamiana, and barley infected with RNAα and -γ plus β.GFP:TGB1 (CP-expressing) RNA with that of RNAα and -γ plus B7.GFP:TGB1 (CP-deficient) RNA. These comparisons revealed no discernible differences in the localization of GFP-TGB1 in the CP⁺ and CP⁻ backgrounds in any of these hosts (data not shown). However, the spread of fluorescence in the GFP-TGB1 CP-deficient foci appeared to be delayed by approximately 24 to 72 h in these hosts. In addition, in C. amaranticolor, smaller lesions appeared in the CP-deficient infections (Fig. 2B). These results suggest that the reductions in the appearance of systemic symptoms in the absence of the CP (B7 background) when the GFP gene is fused to TGB1 may be a consequence of a slightly reduced efficiency of cell-to-cell movement.

TGB1 is associated with membranes.

The subcellular compartmentation of GFP-TGB1 was investigated at higher resolution by inoculating BY-2 protoplasts with wild-type α and γ RNAs plus the β.GFP:TGB1 RNA. GFP fluorescence first appeared 19 to 21 h p.i. and was localized to membranes around the nucleus (Fig. 4A, C, and D) and to discrete foci at the plasma membrane (Fig. 4B, C, and D). In the subsequent 18 h, little alteration in the localization of fluorescence was observed. To confirm that the GFP fusion had no effect on the localization of TGB1, immunofluorescence of BSMV-infected BY-2 protoplasts was evaluated with antiserum raised against the TGB1 protein. At 18 h p.i., antibody fluorescence against the unfused TGB1 protein was primarily localized to membranes surrounding the nuclei (Fig. 4E), and by 24 h p.i., bright spots of antibody fluorescence were also evident at the plasma membrane (Fig. 4F). The foci present at the plasma membrane were slightly more diffuse in protoplasts following immunofluorescence than was GFP fluorescence observed in GFP-TGB1-infected protoplasts. This was due to fixation of the protoplasts prior to immunofluorescence in contrast to nonfixed GFP-fluorescing protoplasts. The punctate foci at the plasma membrane appeared not to be associated with the presence of remnants of the cell wall or with newly deposited cell wall components because calcofluor white staining did not reveal any residual cell wall material (data not shown). From these results, we conclude that the subcellular localization patterns of the native TGB1 protein and the GFP-TGB1 fusion protein are indistinguishable.

FIG. 4.

Subcellular localization of GFP-TGB1 and γb-GFP in BY-2 tobacco protoplasts. The protoplasts were transfected with BSMV RNAs and examined at 18 to 24 h p.i. (A to D) Protoplasts transfected with α, β.GFP:TGB1, and γ RNAs and examined at 22 h p.i. Protoplasts under bright-field illumination (A) and GFP fluorescence (B, C, and D) are shown. Note that panel B is focused on the cell surface and the GFP fluorescence is localized to punctate foci associated with the plasma membrane, while panels C and D are focused on the interior of the cell and GFP fluorescence is accentuated at perinuclear membranes. (E and F) Protoplasts transfected with α, β, and γ RNAs and fixed at 18 (E) or 24 (F) h p.i. and then subjected to immunofluorescence with antiserum raised to the TGB1 protein. (G) Protoplasts transfected with α, β, and γb-GFP RNAs and examined at 22 h p.i. (H) Protoplasts transfected with PVX KDEL-GFP at 24 h p.i. (I and J) Protoplasts transfected with α, β.GFP:TGB1, and γ RNAs and treated with rhodamine B hexyl ester chloride R6 at 22 h p.i. and examined using the fluorescein isothiocyanate (FITC) channel (I) or the rhodamine channel (J). (K and L) Protoplasts transfected with α, β.GFP:TGB1, and γ RNAs, treated with ER-tracker blue-white DPX at 22 h p.i., and examined using the FITC channel (K) or the DAPI (4′,6′-diamidino-2-phenylindole) channel (L). Bar = 30 μm.

The localization of TGB1 in BY-2 protoplasts differed markedly from the localization of the γb-GFP fusion protein following transfection with α- and γb-GFP RNAs. In γb-GFP-infected protoplasts, GFP fluorescence was first observed at approximately 16 h p.i. and appeared to be generally distributed throughout the cytoplasm with no evidence of membrane associations (Fig. 4G). In particular, the fluorescence patterns were diffuse and punctate foci were not observed at the plasma membrane. Because the symptom phenotype and the fluorescence observed in plants infected with B7.GFP:TGB1 (no CP expression) was delayed, the subcellular localization of GFP-TGB1 was also examined in the B7.GFP:TGB1 derivative. Failure to express the CP appeared to have no effect on the localization of GFP-TGB1 to perinuclear membranes or on the appearance of punctate foci at the plasma membranes of BY-2 protoplasts (data not shown), although the timing of the fluorescence was delayed between 3 and 4 h. This delay in fluorescence does not appear to be due to a decreased level of replication, because the GFP-TGB1 RNA abundance was not altered in the presence or absence of the CP (Fig. 1B).

The interactions of TGB1 with cytoplasmic structures in infected protoplasts were assessed by use of cytochalasin D to disrupt actin filaments and treatment at 4°C to disassemble microtubules. At 22 and 26 h p.i., cytochalasin D and cold treatments failed to affect the fluorescence patterns observed with untreated protoplasts (data not shown). Evidence for localization of GFP-TGB1 to the endoplasmic reticulum (ER) in BY-2 protoplasts was provided by comparisons with the distribution patterns of three ER markers. In comparative observations, the KDEL ER localization sequence fused to GFP (KDEL-GFP) expressed from PVX (provided by D. Baulcombe) overlapped with the GFP-TGB1 signals (Fig. 4C, D, and H). Additional tests were conducted by treatment with the dye rhodamine B hexyl ester chloride R6 or ER-tracker blue-white DPX, each of which localizes to ER membranes. Again, the fluorescence of both the rhodamine B hexyl ester chloride R6 and the ER-tracker blue-white DPX marker dyes coincided with GFP-TGB1 in the protoplasts (Fig. 4I to L). Taken together, these results provide substantial evidence that GFP-TGB1 forms prominent associations with the ER and that cytoskeletal interactions, if present, are relatively minor.

Subcellular localization of GFP-TGB1 is affected by mutations in the TGB1 helicase motif.

We previously noted that site-specific mutations in the helicase domain of TGB1 prohibited cell-to-cell movement of BSMV in planta (27). To ascertain whether subcellular localization of GFP-TGB1 containing each of these helicase mutations is altered, RNAα and -γ and the individual β.GFP:TGB1 derivatives containing mutations (M1, M2, M3, M4, M5, M6, M7, and M5/M7 [Fig. 5A]) in each of the six conserved motifs of the helicase domain were used to transfect BY-2 protoplasts. Each of these mutants exhibited pronounced alterations in the subcellular localization of GFP-TGB1. In most cases, fluorescence of the mutant fusion proteins was associated with the plasma membrane and ER membranes surrounding the nucleus, but the proportion of protoplasts exhibiting punctate foci at the plasma membrane declined precipitously (eightfold or more) compared to wild-type GFP-TGB1 (Table 1; Fig. 5B to D). In each case, full-length GFP-TGB1 fusion protein was found to be expressed in these protoplasts by Western blot analyses using antiserum raised against GFP (Fig. 5E) or TGB1 (data not shown).

FIG. 5.

Effects of mutations in the TGB1 helicase domain and the TGB2 and TGB3 proteins on subcellular localization of GFP-TGB1 in BY-2 protoplasts. (A) Amino acid sequence of the helicase domain in TGB1. The overlined amino acids labeled I to VI indicate the conserved regions that define the helicase motif (24). M1 to M7 indicate the positions of site-specific mutations engineered into these regions. (B to D) α and γ RNAs and a wild-type or mutant GFP:TGB1 RNA were transfected into BY-2 protoplasts and examined at 21 to 24 h p.i. (B) Wild-type GFP-TGB1 showing punctate foci; (C and D) shift in fluorescence of GFP-TGB1 helicase mutants M5 (C) or M5/M7 (D) away from punctate foci at the plasma membrane. Similar shifts were noted with the M1, -2, -3, -4, -6, and -7 mutants. (E) Immunoblot analyses of proteins from mock-transfected protoplasts (M) or protoplasts transfected with α and γ RNAs and a wild-type (wt) or a mutant GFP-TGB1 RNA (M1, -2, -3, -4, -5, -6, -7, and M5/M7) at 24 h p.i. Immunoblots were probed with antiserum raised to GFP. (F to H) RNAs α and γ and GFP-TGB1 expressing TGB2 and TGB3 (F and G) or GFP-TGB1 that did not express the TGB2 and TGB3 proteins (H) were transfected into BY-2 protoplasts and examined at 21 to 24 h p.i. Bar = 30 μm.

TABLE 1.

Effects of helicase mutations on subcellular localization of GFP-TGB1

TGB1 derivative	Percentage of protoplasts exhibiting GFP fluorescencea
	ER and PM foci	ER and PM, no PM foci	ER only	Pronounced cytoplasmic
WT	79 ± 8.9	21 ± 8.9	0	0
M1	3 ± 3.5	68 ± 7.7	25 ± 9.0	4 ± 3.0
M2	7 ± 2.4	74 ± 10.2	19 ± 12.5	0
M3	5 ± 3.7	61 ± 7.7	28 ± 11.4	6 ± 4.3
M4	11 ± 4.5	50 ± 8.6	23 ± 5.7	16 ± 7.8
M5	3 ± 2.1	79 ± 9.0	13 ± 11.5	5 ± 4.0
M6	12 ± 6.9	70 ± 10.8	12 ± 5.0	6 ± 0.8
M7	2 ± 1.6	74 ± 17.0	16 ± 19.0	8 ± 3.7
M5/M7	0	79 ± 13.5	15 ± 15.9	6 ± 4.9

^aIn each experiment, 100 BY-2 protoplasts were observed at 23 to 26 h posttransfection. The results shown are the mean and the standard deviations of three separate experiments and are shown for sites of subcellular localization. ER and PM foci, protoplasts with GFP fluorescence associated with ER membranes and/or punctate foci present at the plasma membrane; ER and PM, no PM foci, protoplasts showing diffuse GFP fluorescence associated with ER and plasma membranes but no punctate foci at the plasma membrane; ER only, protoplasts with GFP fluorescence present only at the ER membranes; pronounced cytoplasmic, protoplasts with diffuse GFP fluorescence in the cytoplasm, the ER, and the plasma membranes and only a few punctate foci at the plasma membrane. WT, wild type. 

TGB2 and TGB3 facilitate localization of TGB1 to the plasma membrane.

To investigate the effects of the TGB2 and TGB3 proteins on the subcellular localization of GFP-TGB1, RNAα and -γ plus β.GFP:TGB1 RNA unable to express TGB2 and TGB3 were transfected into BY-2 protoplasts, and the localization of GFP-TGB1 was monitored by fluorescence microscopy at 22 to 24 h. Protoplasts transfected with the TGB2- and TGB3-deficient derivative exhibited a diffuse GFP-TGB1 fluorescence associated with the perinuclear ER and a drastic shift of fluorescence away from the plasma membrane (Table 2; Fig. 5F, G, and H). Similar results were obtained with derivatives expressing only TGB2 or TGB3 (data not shown).

TABLE 2.

Effects of TGB2 and TGB3 on subcellular localization of GFP-TGB1

Genotype	Percentage of protoplasts exhibiting GFP fluorescencea
	ER and PM foci	ER and PM, no PM foci	ER only	Pronounced cytoplasmic
WT	94 ± 1.2	0	6 ± 1.2	0
ΔTGB2-TGB3	5 ± 3.4	0	95 ± 3.4	0

^aIn each experiment, 100 BY-2 protoplasts were observed at 23 to 26 h posttransfection. The results shown are the mean and the standard deviations of three separate experiments and are given for sites of subcellular localization. ER and PM foci, GFP fluorescence associated with ER membranes and/or punctate foci present at the plasma membrane; ER and PM, no PM foci, diffuse GFP fluorescence associated with ER and plasma membranes but no punctate foci present at the plasma membrane; ER only, GFP fluorescence associated only with ER membranes; pronounced cytoplasmic, diffuse GFP fluorescence in the cytoplasm associated with ER and plasma membranes and a few punctate foci at the plasma membrane. 

DISCUSSION

To examine the initial phases of infection of BSMV in monocots and dicots, we have constructed GFP fusions with TGB1 and γb (27). These fusions each provide reliable indicators of the early stages of virus movement, and the GFP-TGB1 results verify and extend the differences shown by γb-GFP during infections of dicot and monocot hosts. In dicot hosts, infections typically spread as circular-to-oval foci through several parenchyma cells separating the reticulate vascular tissue. In contrast, in the monocot host, single cells exhibiting green fluorescence were extremely rare, and infections quickly reached the vascular system and spread throughout the plant.

In dicots, GFP-TGB1 fluorescence was observed by 24 h p.i. in the ER and pronounced punctate foci formed along the cell walls. GFP-TGB1 also exhibited a temporal pattern of expression in dicots, as has been reported for the 42-kDa BNYVV TGB1 protein (15) and the TMV 30-kDa protein (42). This differs markedly from the constitutive cytosolic localization of γb-GFP (27). The temporal expression of the TMV movement protein, P30-GFP, in N. benthamiana has been previously attributed to both enhanced protein degradation and decreased expression near the centers of the foci (42, 59). We have no direct evidence to determine whether TGB1 turns over more rapidly than γb following the initial burst of synthesis. Previous time course studies of RNA synthesis in barley protoplasts suggest that the temporal βb expression may in part be a consequence of early expression of sgRNAβ1 (67; cf. Fig. 1). In this case, sgRNAβ1 appeared to reach maximum abundance by 12 h p.i., whereas the genomic RNAs and sgRNAγ continued to increase in abundance throughout the 24-h period of analysis.

Infections in barley differed markedly from those of the dicot hosts in the timing of GFP-TGB1 fluorescence and in the patterns of expression in infected cells. GFP-TGB1 fluorescence was not evident until 72 h p.i., and no punctate foci were detected in the mesophyll or epidermal cells of infected barley leaves. GFP-TGB1 fluorescence also appeared to be relatively consistent throughout the infected tissue and could be observed for at least 7 days p.i. This suggests that GFP-TGB1 has different patterns of expression and localization in monocot and dicot cells during the initial stages of infection. However, additional experiments are required to determine whether these differences result from primary adaptations to barriers to cell-to-cell movement presented by monocots and dicots or whether they represent indirect host-specific events of little consequence to cell-to-cell movement per se.

Interestingly, the BSMV 58-kDa TGB1 protein and the TGB1 proteins from BNYVV (42 kDa), PCV (51 kDa), and PMTV (51 kDa) have variable amino-terminal extensions that are absent in the smaller (24- to 28-kDa) TGB1 proteins of the potexviruses. At present, the roles of these amino-terminal extensions of the hordeivirus, benyvirus, pecluvirus, and pomovirus TGB1 proteins is unclear. In particular, our lack of knowledge of the significance of these regions is highlighted by findings that in vitro RNA binding by the 42-kDa TGB1 protein of BNYVV (3) maps to the first 24 amino acids but, surprisingly, that site-specific mutations of the basic residues in this region fail to disrupt cell-to-cell movement (15). In contrast, mutagenesis experiments indicate that several regions of the BSMV TGB1 protein function in RNA binding. The amino-terminal region of the TGB1 protein contains elements that contribute to double-stranded RNA versus ssRNA binding in vitro (13), but the functions of these regions in RNA binding in vivo have not been evaluated. However, several deletions affecting RNA binding also compromise cell-to-cell movement (27). The amino-terminal extensions of the larger TGB1 proteins may be able to compensate in part for requirements for the CP for cell-to-cell movement of those viruses containing a smaller TGB1 protein. Analysis of the TGB1 genes of benyviruses, carlaviruses, hordeiviruses, pecluviruses, pomoviruses, and potexviruses suggests that these genes may have evolved from a common progenitor, based on the conservation of the helicase domain present in these proteins (63). However, it is also possible that during the course of evolution, the TGB1 proteins of BSMV, BNYVV, PCV, and PMTV may have acquired amino-terminal functions from diverse sources that permitted them to dispense with a CP requirement for cell-to-cell movement. Alternatively, the larger TGB1 proteins may have retained some functions that the potexvirus TGB1 proteins have lost.

The 42-kDa BNYVV (15) and 51-kDa PCV (16) TGB1 proteins are similar to the BSMV TGB1 protein in that they form punctate foci at the cell walls of infected dicot leaf cells. Specifically, the BNYVV (15) and PCV (16) TGB1 proteins localize to plasmodesmata, while the BSMV TGB1 foci are often present as paired foci that appear to traverse the walls of adjacent cells. This localization pattern contrasts with that of the smaller TGB1 proteins of the potexviruses Bamboo mosaic virus (8), Foxtail mosaic virus (50), and PVX (10) that do not localize to the cell walls in infected cells. In this regard, the potexvirus TGB1 proteins exhibit subcellular localization patterns more typical of BSMV in barley. From these results, it is evident that the TGB1-encoded proteins of hordeiviruses, benyviruses, pecluviruses, and the potexviruses differ substantially in several aspects of their function and interactions. However, the host-specific differences in cell wall associations of the BSMV TGB1 protein in monocots versus dicots complicates straightforward interpretation of results based on cytological analyses in a single host. Therefore, it is evident that more detailed analyses of TGB activities in a variety of hosts are needed to elucidate nuances of the movement processes of these viruses.

The patterns of localization of GFP-TGB1 in cells of infected plants and in protoplasts were similar in that GFP-TGB1 formed punctate foci at the cell walls and at the plasma membrane, respectively. These foci could possibly have been associated with remnants of the cortical ER at sites where the cell wall was attached. However, the foci apparently do not colocalize with cell wall fragments because calcofluor white staining failed to reveal such remnants in the vast preponderance of infected cells. The protoplast localization studies also revealed an association of TGB1 with ER membranes located in close proximity to the nucleus in the absence of TGB2 and TGB3. However, our examination of the TGB1 sequence failed to reveal known ER signal sequences. Even though BSMV TGB1 does not form substantial interactions with cytoskeletal components, its membrane associations appear to be similar to those of the TMV P30 movement protein, which behaves as an integral ER membrane protein (21, 48) despite the lack of known ER retention sequences. Although the exact mechanisms of TMV P30 and BSMV TGB1 membrane associations have not yet been resolved, it is possible that TGB1 ER localization may be mediated by interactions with host proteins that serve to target the movement protein. In this regard, a host transmembrane protein, TOM1, has been identified in Arabidopsis thaliana that interacts with the helicase domain of the TMV replicase, suggesting that this protein may function as an ER membrane anchor for the replicase (64). Although our individual helicase mutations argue against direct helicase domain interactions, multiple host contact sites might exist within the protein.

We have previously demonstrated that site-specific mutations introduced into each of the six conserved helicase motifs of the TGB1 protein abrogate cell-to-cell movement (27). Therefore, even though we were unable to detect helicase activity in vitro using purified TGB1 protein (13), each conserved element in the helicase domain is critical for cell-to-cell movement (27). Moreover, in infected protoplasts, a series of mutations in the TGB1 helicase motif each shifted the subcellular localization of GFP-TGB1 away from the plasma membrane. The mutant GFP-TGB1 proteins showed a much lower proportion of punctate foci at the plasma membrane than the wild-type protein, but the mutants still exhibited an intense membrane fluorescence. These results suggest that the helicase motif may also function to facilitate plasmodesmal interactions with TGB2 and TGB3, or with host proteins required for plasmodesma associations. The fact that both TGB2 and TGB3 are required for formation of punctate plasma membrane foci favors a model wherein TGB2, TGB3, or both proteins interact directly with TGB1. Our ongoing studies of the interactions of these proteins with a ribonucleoprotein complex formed in planta (D. M. Lawrence, J. Yu, and A. O. Jackson, unpublished data) may provide additional details to clarify this model.

Several lines of evidence also indicate that interactions of the TGB proteins of other viruses are required for cell-to-cell movement. For example, like BSMV, the TGB1 protein of BNYVV fails to form punctate foci at the cell walls of infected Chenopodium quinoa leaf cells in the absence of the TGB2 and TGB3 proteins (15). The PCV TGB1 protein is also associated with plasmodesmata in infected cells, but when TGB1 is expressed in transgenic plants, it is not localized to plasmodesmata until after virus infection (16). The TGB1 protein of PVX is able to move from cell to cell when expressed alone but is restricted to single cells when expressed in the presence of TGB2 and TGB3 or the CP, suggesting that the TGB functions may be facilitated by direct interactions (65). Interestingly, the relative levels of expression of the TGB2 and TGB3 proteins of BNYVV also appear to be important for cell-to-cell movement (4), and these results may provide evidence for interactions of TGB2 and TGB3. However, transient-expression experiments in epidermal leaf cells with the hordeivirus Poa semilatent virus-encoded TGB2 and TGB3 proteins have been interpreted to suggest that these two proteins do not form complexes, even though TGB3 assists in the targeting of TGB2 (58). Irrespective of the mechanistic aspects of movement, the available results suggest the TGB proteins can function with heterologous viruses. For example, it has been shown that limited cell-to-cell movement functions of the BSMV TGB can be replaced by the Poa semilatent virus TGB (57), and similar experiments show that the BNYVV TGB can be replaced by the PCV TGB (26). However, in the case of BNYVV, the individual TGB proteins could not be replaced by their respective PCV homologs. Currently, we are conducting experiments to determine how interactions among the BSMV TGB1, -2, and -3 proteins function in the formation of competent cell-to-cell movement complexes.