- American Society of Plant Biologists
Abstract
Many plant viruses exploit a conserved group of proteins known as the triple gene block (TGB) for cell-to-cell movement. Here, we investigated the interaction of two TGB proteins (TGB2 and TGB3) of Potato mop-top virus (PMTV), with components of the secretory and endocytic pathways when expressed as N-terminal fusions to green fluorescent protein or monomeric red fluorescent protein (mRFP). Our studies revealed that fluorophore-labeled TGB2 and TGB3 showed an early association with the endoplasmic reticulum (ER) and colocalized in motile granules that used the ER-actin network for intracellular movement. Both proteins increased the size exclusion limit of plasmodesmata, and TGB3 accumulated at plasmodesmata in the absence of TGB2. TGB3 contains a putative Tyr-based sorting motif, mutations in which abolished ER localization and plasmodesmatal targeting. Later in the expression cycle, both fusion proteins were incorporated into vesicular structures. TGB2 associated with these structures on its own, but TGB3 could not be incorporated into the vesicles in the absence of TGB2. Moreover, in addition to localization to the ER and motile granules, mRFP-TGB3 was incorporated into vesicles when expressed in PMTV-infected epidermal cells, indicating recruitment by virus-expressed TGB2. The TGB fusion protein-containing vesicles were labeled with FM4-64, a marker for plasma membrane internalization and components of the endocytic pathway. TGB2 also colocalized in vesicles with Ara7, a Rab5 ortholog that marks the early endosome. Protein interaction analysis revealed that recombinant TGB2 interacted with a tobacco protein belonging to the highly conserved RME-8 family of J-domain chaperones, shown to be essential for endocytic trafficking in Caenorhabditis elegans and Drosophila melanogaster. Collectively, the data indicate the involvement of the endocytic pathway in viral intracellular movement, the implications of which are discussed.
INTRODUCTION
To establish a successful infection, plant viruses move from cell to cell into the vascular system, where they are systemically translocated throughout the plant. Cell-to-cell movement involves passage through plasmodesmata (PD), the small pores that interconnect cells (reviewed in Roberts and Oparka, 2003). It is now clear that plant viruses have evolved different strategies for cell-to-cell movement that involve one or more virus-encoded movement proteins (reviewed in Carrington et al., 1996; Morozov and Solovyev, 2003). In this report, we focus on the conserved group of three partially overlapping reading frames encoding the triple gene block (TGB) of movement proteins.
The TGB module is a conserved genetic element found in several different viral genera that infect both monocotyledonous and dicotyledonous plants. Two groups of TGB-containing plant viruses have been distinguished (Morozov and Solovyev, 2003): the hordei-like (Group 1; typified by the Hordeivirus Barley stripe mosaic virus; BSMV) and the potex-like (Group 2; typified by the Potexvirus Potato virus X; PVX). In this work, we examine the TGB of the soil-borne Pomovirus Potato mop-top virus (PMTV), a Group 1 TGB virus that is transmitted by the plasmodiophorid Spongospora subterranea, a protozoan parasite that invades root hairs and potato tuber epidermal cells (Braselton, 1992, 1995; Arif et al., 1995).
The first of the TGB proteins (TGB1) is a sequence nonspecific RNA binding protein that contains seven NTPase motifs typical of a viral helicase. In Group 1 viruses, TGB1 contains an additional N-terminal domain of variable mass (∼20 to 40 kD), whereas in Group 2 viruses, TGB1 has a molecular mass of ∼25 kD, encoding only the helicase domain. The sequence of the second protein (TGB2; ∼12 to 14 kD) of both groups contains two predicted transmembrane helices. The N and C termini are predicted to be located in the cytoplasm, and there is a conserved hydrophilic motif (G-G-x-Y-R/K-D-G) in the intervening loop. The TGB2 of PMTV was shown previously to bind RNA in Northwestern gel blots (Cowan et al., 2002).
Unlike TGB2, TGB3 proteins are more heterogeneous both in size and sequence. For example, the TGB3 proteins of hordeiviruses and pomoviruses have a mass of ∼17 to 21 kD, and the amino acid sequences contain two predicted transmembrane helical domains (with N and C termini inside), a conserved Cys residue motif (H-x3-C-x-C-x2-C) in the N terminus, and a conserved motif (Y-Q-D-L-N) in the intervening loop. Potexvirus TGB3s are smaller (∼7 to 8 kD), and the sequences contain a single hydrophobic (transmembrane) domain within the C terminus that is predicted to be cytoplasmic.
The TGB proteins are essential for virus movement. A major difference between the two TGB-containing virus groups is the requirement for coat protein (CP) for movement. Group 2 viruses require the CP, whereas in viruses belonging to Group 1, the CP is not required for cell-to-cell movement. In hordeiviruses and pomoviruses, the CP is also dispensable for long-distance movement (Petty and Jackson, 1990; Savenkov et al., 2003). These observations indicate differences in the composition of the ribonucleoprotein (RNP) movement complex between the two groups. Clearly, if the viral CP is dispensable for movement of the Group1 viruses, then the RNP cannot move in the form of intact virions. In the potexvirus group, two models have been proposed, in which movement may occur as intact virions (Santa Cruz et al., 1998) or, alternatively, as a nonvirion RNP complex (Lough et al., 2000). PVX CP does not modify the size exclusion limit (SEL) of PD but is translocated with the infectious material (Santa Cruz et al., 1998). The TGB1 protein of PVX and White clover mosaic virus can move cell to cell independently of the other TGB proteins and is also translocated with viral RNA (Lough et al., 2000; Yang et al., 2000). In biolistic bombardment experiments, in which TGB proteins and CP were transiently expressed in cells to complement the corresponding defective genes in PVX clones, TGB1 and CP rescued mutant virus movement across several cell layers, but TGB2 and TGB3 only facilitated movement of defective virus to the adjacent cells. This indicates that these proteins were not themselves transported between cells, presumably because of their strong association with cellular membranes (Lough et al., 2000).
Recently, PVX TGB2 has been shown to increase the SEL of PD, enabling the passage of free green fluorescent protein (GFP) between infected cells (Tamai and Meshi, 2001). Also, fusions of GFP to PVX TGB2 and TGB3 were shown to move to adjacent cells when the proteins were expressed from plasmids bombarded onto Nicotiana benthamiana leaves (but not in five other host plants tested). The proteins also moved to adjacent cells when bombarded onto transgenic N. tabacum expressing TGB1 (Krishnamurthy et al., 2003).
BSMV (Group 1) TGB1 was associated with perinuclear endoplasmic reticulum (ER) membranes and punctate foci at the plasma membrane when expressed from a BSMV-based vector in tobacco BY-2 protoplasts. In the absence of TGB2 and TGB3, fluorescence was only associated with perinuclear ER and not plasma membranes (Lawrence and Jackson, 2001a). Similarly, punctate spots that colocalized with callose were seen in the walls of cells expressing Beet necrotic yellow vein virus (Group 1) TGB1 in the presence of TGB2 and TGB3 (Erhardt et al., 2000). The punctate spots are thought to be plasmodesmata.
Previously, we studied the TGB proteins of PMTV and described their subcellular localizations when expressed as N-terminal fusions to GFP from a Tobacco mosaic virus (TMV)–based vector (Cowan et al., 2002). In these experiments, both the PMTV GFP-TGB2 and GFP-TGB3 fusions were associated with cellular endomembranes, particularly the ER network and membranes surrounding the nucleus. Fluorescent spots and aggregates were seen to move along the ER and transvacuolar strands, and stationary aggregates were seen at the periphery of the cells. In addition, GFP-TGB3 labeled opposing pairs of fluorescent spots across neighboring cell walls that suggested localization at, or near, plasmodesmata.
The above findings are in general agreement that the movement proteins of several viruses may associate with the endomembrane system (see Oparka, 2004) and are in line with the reported localization of TGB proteins of Poa semilatent hordeivirus (PSLV; Solovyev et al., 2000) and PVX (Krishnamurthy et al., 2003; Mitra et al., 2003), which in transient expression assays showed an association with the ER network or ER-derived structures at the cell periphery (Solovyev et al., 2000; Zamyatnin et al., 2002). Mutations in the predicted transmembrane domains of PVX TGB proteins that disrupt membrane binding inhibited the association with ER and virus movement (Krishnamurthy et al., 2003; Mitra et al., 2003). In studies with PSLV, coexpression of the two proteins revealed that they colocalized in the cell and that TGB3 redirected localization of TGB2 from the ER network to the peripheral bodies (Solovyev et al., 2000). TGB3-assisted targeting to the periphery has also been shown with unrelated proteins, such as PVX TGB2 and the 6-kD movement protein of Beet yellows closterovirus (Solovyev et al., 2000; Zamyatnin et al., 2002). Also, evidence for the association of PSLV TGB3 with plasmodesmata has been obtained from experiments with GFP-TGB3 transgenic plants in which colocalization with callose deposits was observed (Gorshkova et al., 2003). Recent studies have shown that on transient expression of PMTV GFP-TGB1, fluorescence was seen in the cytosol and nuclei, whereas when GFP-TGB1 was coexpressed with TGB2 and TGB3, the fluorescence appeared in peripheral bodies close to the plasma membrane (Zamyatnin et al., 2004). Furthermore, the same authors showed that when GFP-TGB1 was expressed alone, fluorescence was confined to the transfected cell, but when GFP-TGB1 was coexpressed with TGB2 and TGB3, fluorescence moved into the adjacent cells.
It is generally accepted that the TGB1 protein interacts with RNA and is transported, as part of the RNP complex, out of the cell and long distance throughout the plant. The TGB2 and TGB3 proteins facilitate RNP intracellular transport, but the details of this process are not well understood, and the respective roles of these two integral membrane proteins are the focus of this report. Our current hypothesis is that TGB2 and TGB3 act as endomembrane anchors, where they bind viral RNP to direct its transport to plasmodesmata. We have conducted a detailed investigation of the subcellular pathways involved in trafficking of the PMTV TGB2 and TGB3 to the cell periphery. Expression of GFP-tagged and monomeric red fluorescent protein (mRFP)–tagged TGB proteins was examined in wild-type and transgenic plants in which a range of specific subcellular compartments were either stained with specific markers or labeled with GFP. In addition, the interaction of TGB2 and TGB3 with host factors was investigated. We dissect the distinctive roles of each protein and provide several lines of evidence to show that the TGB2 and TGB3 proteins of PMTV interact with components of the endocytic pathway. The data are discussed in the light of the emerging evidence for endocytosis in plants.
RESULTS
TGB2 and TGB3 Fusion Proteins Show an Early Association with the ER
In biolistic bombardment experiments, conducted with plasmids expressing TGB2 or TGB3 fusion proteins under the control of the 35S promoter (P35S) or when expressed from TMV vectors, both fusion proteins showed an early association with the cortical ER, whether introduced independently or together. Examination of transgenic plants that express GFP in the ER (erGFP) bombarded with P35S-mRFP-TGB3 revealed that fluorescence was seen first in the cortical ER, which became labeled uniformly. Figures 1A to 1C show the appearance of the cortical ER dual labeled with mRFP-TGB3 and GFP 24 to 48 h after bombardment. In addition to labeling the ER network, P35S-expressed GFP-TGB3 became associated with motile punctate structures located at or on the ER membranes 2 to 3 d after bombardment (Figure 1D).
Expression of TGB2 and TGB3 Fluorescent Fusion Proteins in Epidermal Cells.
(A) to (C) Transient expression (P35S) of mRFP-TGB3 on transgenic N. benthamiana expressing GFP in the ER showing an early association of mRFP-TGB3 with ER. Bar = 10 μm for (A) to (C).
(A) mRFP-TGB3 labels the cortical ER at 1 to 2 d post-bombardment (dpb) (red channel).
(B) Same epidermal cell as in (A), showing the ER network in this transgenic ER-GFP plant (green channel).
(C) Overlay of images (A) and (B). Notice that there is good colocalization of mRFP-TGB3 with the strands of the ER network. mRFP-TGB3 also shows aggregation at certain points of the ER vertices.
(D) Bombardment of P35S-GFP-TGB3 onto epidermal cells of N. benthamiana. Later (2 dpb), GFP-TGB3 appears on small motile bodies that move along the ER network. Bar = 15 μm.
(E) and (F) Bombardment of TMV-GFP-TGB2 onto epidermal cells of N. benthamiana.
(E) GFP-TGB2 shows early association with the ER, producing a uniform labeling of the cortical ER network at 1 to 2 dpb. Bar = 10 μm.
(F) Later in the infection, GFP-TGB2 also appears on small motile bodies that move along the ER network (cf. TGB3 in Figure 1D). Bar = 15 μm.
(G) Cobombardment of P35S-mRFP-TGB2 and P35S-GFP-TGB3 onto epidermal cells of N. benthamiana. Transiently expressed mRFP-TGB2 and GFP-TGB3 colabel the ER network and motile structures, but labeling of the ER network with mRFP-TGB2 was generally patchier than GFP-TGB3 2 dpb. Bar = 10 μm.
(H) to (J) Later in expression, TGB2 labeled a motile vesicle population. All images are in N. benthamiana epidermal cells.
(H) At 2 dpb, transiently expressed (P35S) mRFP-TGB2 labeled a motile vesicle population (arrow). Bar = 20 μm.
(I) One day later, these TGB2-labeled vesicles accumulated around the nuclear envelope (3 dpb). Bar = 10 μm.
(J) Cobombardment of P35S-GFP and P35S-mRFP-TGB2. The expression of free GFP in the cytosol clearly shows that the mRFP-TGB2–labeled vesicles are contained in the cytosol (2 dpb). Bar = 10 μm.
(K) to (M) Cobombardment of P35S-mRFP-TGB2 and P35S-GFP-TGB3 onto an epidermal cell of N. benthamiana at 2 dpb. Bar = 10 μm for (K) to (M).
(K) mRFP-TGB2 labels motile bodies and vesicles (as shown in [H]) (red channel). Arrows point to vesicles.
(L) GFP-TGB3 labels the ER network of the same cell, in addition to the motile structures and vesicles (arrows) (green channel).
(M) Overlay of images (K) and (L) showing colabeling of vesicles (arrows) by both TGB proteins.
(N) to (P) Magnified images showing colabeling of vesicles with mRFP-TGB2 and GFP-TGB3 at 2 dpb. Bar = 10 μm for (N) to (P).
(N) mRFP-TGB2 shows strong labeling of the vesicle membrane (arrow) (red channel).
(O) GFP-TGB3 also shows fluorescence predominantly on the vesicle membrane (arrow) (green channel).
(P) Overlay of images (N) and (P). This shows colocalization of both TGB proteins on the membrane and shows some hot spots on the surface of the vesicles (arrow). Both TGB2 and TGB3 are present in the hotspots.
All of the images are reconstructions of the cell using Z stacking.
Similarly, mRFP-TGB2 or GFP-TGB2 also labeled the ER initially. Figure 1E shows early association of TMV-expressed GFP-TGB2 on ER, and subsequently labeled ER-associated motile bodies similar to those seen with TGB3 were observed (Figure 1F; see Supplemental Figure 1 online). In colocalization experiments, mRFP-TGB2 and GFP-TGB3 showed colocalization to the ER, with the pattern of mRFP-TGB2 labeling generally being more patchy than that of GFP-TGB3 (Figure 1G). In addition, both TGB2 and TGB3 fusion proteins colocalized in the motile granules and labeled ER membranes surrounding the nucleus.
In these experiments, identical results were obtained regardless of whether the fusion proteins were expressed under the control of P35S, or from TMV, or whether the TMV vector had a functional or defective MP, except that infection was confined to a single cell when expressed from the TMV vector with defective MP (see Supplemental Figure 2 online).
Later, TGB2 Fusion Protein Labels Motile Vesicular Structures
A characteristic feature of the expression cycle of cells bombarded with TGB2 fusion proteins was the appearance of a motile population of vesicle-like structures with a size range of ∼0.5 to 4 μm (Figure 1H). Many of these vesicles were associated with the plasma membrane and seemed to bud from it at discrete locations in the periphery of the cell. Many of the vesicles appeared to move by cytoplasmic streaming, and many became closely associated with the nuclear envelope (Figure 1I). To determine whether the labeled vesicles were present within the cytoplasm, or located within the vacuole, P35S-mRFP-TGB2 plasmid DNA was cobombarded into cells along with P35S-GFP as a cytosolic marker. In these cases, the mRFP-TGB2–labeled vesicles were clearly surrounded by cytoplasm (Figure 1J). When the GFP-TGB3 or mRFP-TGB3 was expressed independently from either P35S or the TMV-based vectors, they were never found to label the vesicular structures.
TGB2 Recruits TGB3 to Vesicles
In cobombardment experiments with P35S-GFP-TGB3 and P35S-mRFP-TGB2, the colocalization of the two proteins in the vesicles was clearly visible (Figure 1K to 1M). The GFP-TGB3 seemed to be more evenly distributed throughout the vesicle, whereas mRFP-TGB2 strongly labeled the membrane (Figures 1N to 1P). At high magnifications, the vesicles often appeared to be beaded, containing small hot spots on their surfaces (Figures 1N to 1P).
Expression of TGB Fusion Proteins in PMTV-Infected Cells Reveals Recruitment of mRFP-TGB3 to Vesicles
To investigate whether the subcellular localization of TGB fusion proteins was affected by the presence of a replicating virus, we bombarded P35S-mRFP-TGB2 or P35S-mRFP-TGB3 onto leaves that had been manually inoculated with PMTV preparations. Tissue print immunoassays were done on the inoculated leaves after imaging to confirm that the inoculated leaves had a uniform distribution of virus-infected cells. The presence of virus was also confirmed by RT-PCR. In these experiments, the presence of replicating virus made no difference to the subcellular localization of mRFP-TGB2 (in five independent experiments). By contrast, the presence of replicating virus did affect the subcellular localization of mRFP-TGB3. Red fluorescence was seen as before in ER membranes and motile spots, but also in vesicles, which were never seen in noninfected cells. This result indicates that the virus-expressed TGB2 recruited mRFP-TGB3 to the vesicles (Figure 2A). In these experiments, neither of the fluorescent fusion proteins moved out of the bombarded cells, indicating that neither the presence of viral RNA nor other virus-expressed TGB proteins enabled the intercellular movement of the membrane anchored TGB2 or TGB3.
Expression of TGB3 in PMTV-Infected Cells, Association of TGB2 and TGB3 with Peripheral Structures, and Plasmodesmal Gating.
(A) and (B) Bombardment of P35S-mRFP-TGB3 on leaves infected with PMTV.
(A) In the presence of PMTV, mRFP-TGB3 is seen localized to motile vesicles (arrows) and small structures, as seen when cobombarded with TGB2 (cf. Figure 1M). Image is a reconstruction of the cell using Z stacking. Bar = 15 μm.
(B) In the presence of PMTV, mRFP-TGB3 produces small stationary punctae at the periphery of the bombarded epidermal cell (arrows). Image is a stack of four through the central plane of the cell. Bar = 20 μm.
(C) TMV expressed GFP-TGB3 shows localization to punctae at the periphery of N. benthamiana epidermal cells at 3 dpb. These stationary deposits appeared two to three cell layers behind the leading edge of the infection site. Image is a stack of four through the central plane of the cell. Bar = 10 μm.
(D) to (F) Images of transgenic plants expressing TMV MP fused to GFP (MP-GFP) in epidermal cells infected with TMV-mRFP-TGB3 at 3 dpb. Images are a stack of four through the central plane of the cell. Bar = 15 μm for (D) to (F).
(D) A constitutively expressed TMV MP-GFP localizes to branched plasmodesmata in epidermal cells of transgenic N. tabacum (green channel).
(E) The same epidermal cell showing stationary red fluorescent deposits of mRFP-TGB3 at the cell periphery (red channel).
(F) Overlay of images (D) and (E) showing that many of the stationary mRFP-TGB3 deposits precisely colocalized with plasmodesmata (arrow), whereas other plasmodesmata were only labeled with the TMV MP-GFP (arrowhead).
(G) Transient expression of P35S-mRFP-TGB2 and P35S-GFP-TGB3 after cobombardment into a N. benthamiana epidermal cell. Two to 3 d after bombardment, both proteins colocalize to aggregates at the cell periphery (arrows). Image is a reconstruction of the cell using Z stacking. Bar = 25 μm.
(H) to (L) Plasmolysis experiments of transient P35S-expressed vectors at 3 dpb in onion epidermal cells. Images are a reconstruction of the cell using Z stacking.
(H) In nonplasmolysed cells, mRFP-TGB3 labels fragments of ER (arrow) and punctae at the cell periphery (arrowhead). Bar = 15 μm.
(I) After plasmolysis, mRFP-TGB3 fluorescence became predominantly cytosolic and was retained inside the protoplast as it retracted from the cell wall. Bar = 20 μm for (I) and (J).
(J) Overlay of image in (I) and a false-transmission image showing lack of red fluorescence in the cell wall after plasmolysis.
(K) After plasmolysis, mRFP-TGB2 fluorescence was also predominantly cytosolic and was retained inside the protoplast after retraction from the cell wall. Bar = 15 μm for (K) and (L).
(L) Overlay of image in (K) and a false-transmission image showing lack of red fluorescence in the cell wall.
(M) to (O) Gating of plasmodesmata by mRFP-TGB2 and mRFP-TGB3 at 2 dpb. Images are a reconstruction of the cell using Z stacking.
(M) Cobombardment of P35S-GFP-sporamin and P35S-mRFP-TGB2. mRFP-TGB2 increases the SEL of plasmodesmata allowing GFP-sporamin to move into adjacent cells. Bar = 35 μm.
(N) Cobombardment of P35S-GFP-sporamin and P35S-mRFP-TGB3. mRFP-TGB3 increases the SEL of plasmodesmata allowing GFP-sporamin to move into adjacent cells. Bar = 35 μm.
(O) The majority of cells cobombarded with P35S-GFP-sporamin and P35S-mRFP showed that the green fluorescence was confined to a single cell. Bar = 25 μm.
Fusion Proteins of TGB3, but Not TGB2, Associate with Plasmodesmata
After the labeling of the ER, conspicuous punctae of fluorescent TGB3 were visible at the cell periphery when TGB3 was expressed from TMV vectors and when transiently expressed in PMTV-infected cells (Figures 2B and 2C). Some of these punctae were motile in the peripheral cytoplasm, whereas others were stationary and remained associated with discrete regions of epidermal cell walls. It has been shown that the movement protein (MP) of TMV, when constitutively expressed either alone or as a GFP fusion (MP-GFP) in transgenic plants, localized to branched PD in epidermal cells (Ding et al., 1992; Roberts et al., 2001). To examine whether the stationary structures were located at plasmodesmata, plasmid P35S-mRFP-TGB3 was bombarded onto leaves of transgenic plants expressing TMV MP-GFP (Figures 2D to 2F). These colabeling experiments showed that many of the stationary mRFP-TGB3 deposits were precisely colocalized with plasmodesmata (yellow signal in Figure 2F), whereas other deposits were not. Thus, at least a subpopulation of TGB3 fusion protein appears to become stably associated with plasmodesmata. In addition, unlike the MP of TMV, which can be seen to target plasmodesmata at the leading edge of the infection (Oparka et al., 1997), when GFP-TGB3 was expressed from a TMV-based vector containing functional MP, labeling of plasmodesmata occurred inside the infection site, consistently approximately two to three layers away from the leading edge (data not shown). These results suggest that for PMTV, accumulation at plasmodesmata occurs relatively late in the infection cycle and is not influenced by the presence of the TMV MP.
In experiments in which GFP-TGB2 was expressed from the TMV vector in N. benthamiana epidermal cells, no labeling of plasmodesmata was observed. In experiments where P35S-mRFP-TGB2 was cobombarded with P35S-GFP-TGB3, no association with the punctae representing plasmodesmata was observed, but dual-labeled stationary structures were seen at the cell periphery (Figure 2G).
Plasmolysis Causes Removal of TGB Fusion Proteins from the Cell Periphery
In onion (Allium cepa) epidermal cells expressing mRFP-TGB2 and mRFP-TGB3 fusions, the localization of the fluorescent proteins was similar to that seen for transient or viral expression in N. benthamiana plants. Fragments of ER were seen throughout the cells and in the perinuclear region, and a mobile vesicle population was frequently labeled with the TGB2 fusion protein (data not shown). mRFP-TGB3 was also seen at peripheral punctae, possibly associated with plasmodesmata (Figure 2H). When the cells were incubated in water, there was no alteration in the localization of fluorescence, but there was a distinct alteration in plasmolysed cells that had been incubated in a sugar solution (Figures 2I to 2L). In both mRFP-TGB2 and mRFP-TGB3–expressing cells, fluorescence became predominantly cytosolic after plasmolysis (Figures 2I and 2K) and was retained inside the protoplast as it retracted from the cell wall (Figures 2J and 2L). No fluorescence was maintained at the cell periphery for either protein, indicating that neither TGB fusion protein was embedded within plasmodesmata.
TGB2 and TGB3 Fusions Both Increase the SEL of Plasmodesmata
In biolistic bombardment experiments in which mRFP-TGB2, mRFP-TGB3, or mRFP was coexpressed in the same cell together with GFP-sporamin, both TGB proteins increased the SEL of plasmodesmata, allowing GFP-sporamin to move to adjacent cells (Figures 2M and 2N). The numbers of cells in which the initial bombarded source cell was surrounded by one or more fluorescent cells were recorded (Table 1). In N. benthamiana cells, there were significantly more cells surrounding the target cells expressing mRFP-TGB2 (28%) or mRFP-TGB3 (35%) than for mRFP alone (4%) (Figures 2M to 2O, respectively). The results from N. tabacum also indicated that both TGB proteins increased the plasmodesmatal SEL, but the numbers of cells showing gating were fewer (23% for TGB2 and 18% for TGB3). We never observed movement of mRFP-TGB2 or -TGB3 from the initial target cell in any of these experiments. An analysis of deviance showed that the small difference in the proportion of cell clusters found between the TGB2 and TGB3 fusion protein treatments within a species (i.e., 28 and 35% for N. benthamiana or 23 and 18% for N. tabacum) was not significant, but it confirmed that the difference between the two species was significant. Therefore, the fusion proteins facilitated a greater increase in plasmodesmatal SEL in N. benthamiana epidermal cells than N. tabacum epidermal cells.
Proportion of Cells Showing Passage of GFP-Sporamin (47 kD), Indicating an Increase in Plasmodesmatal SEL (Gating)
mRFP-TGB3 Mutant YQDLN to GQDGN Does Not Localize to ER or Plasmodesmata
Mutation of the conserved putative Tyr-based sorting motif YQDLN in TGB3 to YAAAA did not affect the subcellular localization of this mutant protein fused to mRFP. However, the mutation YQDLN to GQDGN abolished the targeting of the fusion protein to the ER and motile granules (Figure 3A). Moreover, we did not see an association with plasmodesmata in experiments where the mutant P35S-mRFP-TGB3 (YQDLN to GQDGN) was bombarded onto TMV MP-GFP transgenic plants (Figures 3A to 3C). A separate set of gating experiments was done to compare the mRFP-TGB3 wild type with the mRFP-TGB3 mutant (YQDLN to GQDGN). Each plasmid was cobombarded with P35S-GFP-sporamin onto N. benthamiana plants and results obtained from three independent experiments. The results showed that the gating efficiency of the mutant was impaired compared with the wild type; 90/192 (47%) cells showed gating with wild-type construct compared with 44/190 (23%) with the mutated TGB3.
Plasmodesmal and ER Association of TGB3 Is Lost in Mutant TGB3.
Latrunculin treatment destroys actin cables and stops mRFP-TGB3 movement. All of the images are reconstructions of the cells using Z stacking.
(A) to (C) Images of transiently expressed (P35S) mRFP-TGB3 mutant YQDLN to GQDGN (mutTGB3) in transgenic TMV MP-GFP-expressing N. tabacum epidermal cells at 3 dpb. Bar = 30 μm in (A) to (C).
(A) Epidermal cell showing red fluorescence of mRFP-mutTGB3 in the cytosol but no labeling of ER network or plasmodesmata (red channel).
(B) The constitutively expressed TMV GFP-MP fusion localizes to branched plasmodesmata in the same cells (green channel).
(C) Overlay of images (A) and (B) showing that mRFP-mutTGB3 shows no colocalization with plasmodesmata labeled with the TMV MP (arrowhead).
(D) to (F) Latrunculin treatment destroys actin cables and stops mRFP-TGB3 movement.
(D) GFP-hTalin transgenic A. thaliana. Bar = 10 μm.
(E) Leaves of plant in (D) showing actin cables depolymerized after latrunculin treatment. Bar = 10 μm.
(F) Transient expression of mRFP-TGB3 on leaves of (D); mRFP-TGB3 motile granules associate with and move on GFP-labeled filaments (arrowhead). Some motile granules appear to be encircled by actin (arrows). Bar = 10 μm.
(G) to (I) Transient expression of mRFP-TGB3 on leaves of (D) after latrunculin treatment; actin filaments are depolymerized, and there is no visible mRFP-TGB3 in the ER or moving spots. Even when depolymerized, the TGB3 granules are surrounded by pools of actin. Red fluorescence colocalizes to patches of green fluorescence. Green channel (G); red channel (H); overlay (I). Bar = 10 μm.
(J) to (M) No colocalization of mRFP-TGB3 with microtubules and localization is unaffected by oryzalin treatment.
(J) Transgenic N. benthamiana plants expressing a GFP fusion to TUA6 show GFP-labeled microtubules. After bombardment of leaf epidermal cells, transiently expressed (P35S) mRFP-TGB3 shows no colocalization with microtubules. Image taken at 2 dpb. Bar = 20 μm.
(K) to (M) Same plants as in (J) after treatment with oryzalin; note the absence of GFP-labeled microtubules but presence of mRFP-TGB3–labeled ER and motile spots. Green channel (K); red channel (L); overlay (M). Bar = 5 μm.
Chemical Treatments Show That Movement of TGB Proteins Requires the Actin Network
Leaves of transgenic plants expressing GFP targeted to either the ER (erGFP), the actin cytoskeleton (GFP-hTalin), or microtubules (tua-GFP) were bombarded with either P35S-mRFP-TGB2 or P35S-mRFP-TGB3 and subsequently treated with chemical inhibitors or water. In these plants, transiently expressed mRFP-labeled granules (of both TGB2 and TGB3) coaligned with GFP-actin filaments (Figure 3F) and were seen to move at or on the filaments, but no such association was seen with GFP-labeled microtubules (Figure 3J). After treatment with latrunculin, which disrupted the GFP-hTalin–labeled actin filaments (Figures 3D and 3E) and arrested cytoplasmic streaming in both the erGFP and GFP-hTalin plants, the subcellular distribution of the mRFP-TGB proteins changed radically from visible ER labeling and motile spots to patches of red fluorescence (Figures 3G to 3I), and the movement of both the small compact ER-associated granules and the larger vesicles labeled by either mRFP-TGB2 or mRFP-TGB3 was abolished in both plants. By contrast, treatment with oryzalin, which disrupts microtubules, did not inhibit the movement of either the small granules or vesicles, despite the observed degradation of microtubules at the concentrations used (Figure 3K). Furthermore, the subcellular distribution of red fluorescence was still visible on the ER and in motile spots and vesicles (Figures 3K to 3M). Localization of mRFP TGB2/3 proteins on plants expressing a Golgi-GFP marker (sialyl transferase; Boevink et al., 1998) showed that the granular structures that moved along the ER were not colocalized with Golgi bodies (see Supplemental data 3 online). In addition, treatment of these latter plants with 100 μg mL−1 brefeldin A (BFA), a concentration that in our experiments caused the Golgi bodies to be reabsorbed into the ER network (data not shown), showed no effect on the movement of either the mRFP-TGB2 or mRFP-TGB3 granules or vesicles. The subcellular distribution and movement of fluorescent TGB in cells infiltrated with water was similar to that of untreated cells.
The TGB Containing Vesicles Are Derived from the Plasma Membrane
Because many of the TGB2-labeled vesicles were associated with the plasma membrane, we examined whether the vesicle membrane was derived from the plasma membrane by staining the latter with FM4-64 (Ueda et al., 2001; Bolte et al., 2004). Within 30 min of application to N. benthamiana leaves, the FM4-64 predominantly labeled the plasma membrane (Figure 4D), and after 45 min, an internalized vesicle population was apparent (Figure 4E), showing that this population is present in uninfected cells and is not induced by overexpression of heterologous proteins. Subsequently, N. benthamiana cells expressing GFP-TGB2 were treated with FM4-64, and after 1 to 3 h, dye-labeled vesicles were observed that also showed discrete labeling with GFP-TGB2 (Figures 4A to 4C). Noticeably, not all FM4-64–labeled vesicles contained TGB2 fusion protein, suggesting that only a subpopulation of vesicles contained the internalized TGB2 protein. FM4-64 staining of cells expressing GFP-TGB2 revealed highly dynamic, tubular structures from which the colabeled vesicles were derived (data not shown). These complex structures are similar to those described for the early endosome compartment found in animal cells (Gruenberg, 2001).
TGB2 Associates with Endocytic Vesicles and a Model for Intracellular Movement of PMTV TGB Proteins.
(A) to (C) Colabeling of vesicles with TMV-expressed GFP-TGB2 and FM4-64 in an epidermal cell of N. benthamiana at 2 dpb. Images are a reconstruction of the cell using Z stacking. Bar = 10 μm for (A) to (C).
(A) Two hours after infiltration, FM4-64 labels both the plasma membrane and a population of vesicles (arrows) (red channel).
(B) GFP-TGB2 localizes to the ER and vesicles (arrow) in the same cell as shown in (A) (green channel).
(C) Overlay of images (A) and (B) showing the colocalization of FM4-64 and GFP-TGB2 in the membrane of the vesicle (arrow).
(D) and (E) Labeling of membranes in an epidermal cell of N. benthamiana using the endosomal stain FM4-64.
(D) Half an hour after infiltration, FM4-64 labels the plasma membrane. Image is a stack through the central plane of the cell. Bar =10 μm.
(E) At 45 min after infiltration, vesicles ranging between 0.5 and 4 μm in diameter bud off the plasma membrane and are seen moving through the cytosol. Image is a reconstruction of the cell using Z stacking. Bar = 10 μm.
(F) In epidermal cells of N. benthamiana, transiently expressed (P35S) GFP-Ara7 labels the plasma membrane and a population of variously sized vesicles (arrows) (2 dpb). Image is a reconstruction of the cell using Z stacking. Bar = 20 μm.
(G) to (J) Colabeling of vesicles with mRFP-TGB2 and GFP-Ara7 in epidermal cells of N. benthamiana at 2 dpb. Bar = 10 μm for (G) to (I), which are stacked images through the central plane.
(G) GFP-Ara7 labels vesicles throughout the cytoplasm of the cell, seen here near the cell periphery (arrow) (green channel).
(H) mRFP-TGB2 labels the same vesicle (arrow) and also the plasma membrane (arrowhead) (red channel).
(I) Overlay of images (G) and (H) showing colabeling of the membrane of the vesicles (arrow).
(J) mRFP-TGB2 shows labeling of the nuclear envelope (arrow), whereas GFP-Ara7 labels a population of vesicles that cluster around the nucleus. Image is a reconstruction of the cell using Z stacking. Bar = 5 μm.
(K) and (L) Colabeling of membranes with GFP-Ara7 and FM4-64 in N. benthamiana epidermal cells at 2 dpb (stacked images through the central plane).
(K) The majority of vesicles stained with FM4-64 are also labeled with GFP-Ara7 (arrows). On some occasions, large pleiomorphic vesicles were observed where one vesicle contained multiple smaller vesicles (arrowhead). Bar = 10 μm.
(L) A higher magnification image showing the double-labeled vesicles seen in (K) (arrows). The inset gives a higher magnification image of the pleiomorphic vesicles (seen in [K]); where these were observed, the membranes of both the large enclosing vesicle and smaller internal vesicles were colabeled with FM4-64 and GFP-Ara7. Bar = 10 μm.
(M) Model for intracellular movement of PMTV TGB proteins. Early in the infection cycle, TGB2 (shown as red ovals) and TGB3 (green ovals) associate with the endomembrane system, including the perinuclear membranes. Both proteins label the membranes of the ER and colocalize to small motile granules (orange circles) that move along the ER network, using the actin cytoskeleton for motility. TGB3 contains the information that determines transport specificity and targets the granules to plasmodesmata. The granules dock at the neck region of plasmodesmata, and TGB2, which contains an insertion tag, mediates fusion with the plasma membrane, leaving the TGBs inserted in the plasma membrane at the neck of the plasmodesmata. The viral RNP complex interacts with the plasmodesmal trafficking machinery to permit selective trafficking of the viral RNP. TGB2 contains the signal for recycling of the TGB2/3 complex back, via the plasma membrane–derived vesicles (yellow circles), to the perinuclear membranes using the endocytic pathway. In this way, TGB2 and TGB3-containing granules, lacking the viral RNP, return to the place of viral RNA synthesis to be recycled for future transport processes.
Colocalization of TGB2 with Ara7 (Rab5), a Marker for the Early Endosome, Reveals a Role for the Endocytic Pathway in Retrieving Viral Proteins
To obtain further evidence that the TGB fusion protein–containing vesicles were a component of the plant endocytic pathway, we bombarded a GFP fusion of the Rab protein Ara7, with or without mRFP-TGB2, into epidermal cells. In the absence of any other bombarded proteins, GFP-Ara7 labeled a distinct population of motile vesicular compartments (Figure 4F; see also Ueda et al., 2001). When cobombarded with P35S-mRFP-TGB2, some but not all of the vesicles were shown to contain both mRFP-TGB2 and GFP-Ara7 (Figures 4G to 4I), and some labeling of the plasma membrane by mRFP-TGB2 is also visible. As in previous experiments (see Figure 1I), many of the vesicles were delivered to the vicinity of the nuclear membrane, which showed conspicuous labeling with mRFP-TGB2 but not GFP-Ara7 (Figure 4J). Next, we bombarded cells with GFP-Ara7, followed by FM4-64 staining of the plasma membrane. After application of FM4-64, a distinct vesicle population was observed that showed precise colocalization of GFP-Ara7 and FM4-64 (Figures 4K and 4L). At high magnifications, these vesicles were seen to be highly pleiomorphic and often contained a subpopulation of Ara7/FM4-64–labeled vesicles (Figure 3L, inset). The above data confirm that TGB2 fusion protein does not induce the formation of vesicles in bombarded cells, but rather becomes localized to a pre-existing population of Ara7-positive vesicles.
TGB2 Interacts with a Tobacco-DnaJ Protein Belonging to the Receptor-Mediated Endocytosis-8 Family of Endosomal Trafficking Proteins
To identify host factors that might interact with the TGB proteins, we searched a library of tobacco leaf proteins. The proteins were expressed from colonies induced on replica plates, and in vitro interactions with purified recombinant TGB2 were analyzed using overlay blots. Approximately 105 library clones were screened on the filters. Very few clones displaying strong interactions were visible, indicating specificity of binding. The positive clones were rescued from the noninduced replica plates, and plasmid DNA from five clones was sequenced. Four of the five sequences were identical, and BLAST searches (Altschul et al., 1997) revealed significant sequence identity (E-value = 5e-90) to a DnaJ domain–containing Arabidopsis thaliana protein (GenBank accession number AAC32237) belonging to the Receptor-Mediated Endocytosis-8 (RME-8) family (Zhang et al., 2001) of proteins (Figure 5). The other sequence shared sequence identity to an unknown protein from pea (Pisum sativum) (data not shown).
Amino Acid Sequence Alignment of a TGB2-Interacting N. tabacum Protein (Ntp1) with the RME-8 Protein (AAC3227.1) from Arabidopsis.
Identical sequences are underlined.
DISCUSSION
Previous studies have shown the essential requirement of each of the TGB proteins in virus movement. TGB1 is an RNA binding protein and functions in both viral cell-to-cell and long distance movement, whereas the primary functions of TGB2 and TGB3 are intracellular movement to facilitate passage of the viral RNP complex to and through plasmodesmata (Lough et al., 1998, 2000; Santa Cruz et al., 1998; Erhardt et al., 2000; Lawrence and Jackson, 2001a, 2001b; Zamyatnin et al., 2004). Previous results also suggest that cell-to-cell movement may be regulated by the relative amounts of TGB2 and TGB3 expressed in the cell (Bleykasten-Grosshans et al., 1997; Yang et al., 2000; Tamai and Meshi, 2001). The genetic arrangement where TGB proteins are expressed from overlapping reading frames may well facilitate the regulation of translation, but this arrangement complicates investigations of the intracellular movement of the TGB2 and TGB3 when expressed from the homologous virus. Moreover, TGB proteins are expressed at low levels and are not readily detectable in virus-infected plants. Therefore, in several studies, the role and functions of TGB2 and TGB3 proteins have been investigated using microinjection or biolistic bombardment to introduce plasmids for transient expression of fluorescent fusion proteins (Solovyev et al., 2000; Zamyatnin et al., 2002, 2004; Krishnamurthy et al., 2003; Mitra et al., 2003). In addition, expression of fusion proteins from viral vectors has provided important information on the function and localization of viral and plant proteins (Blackman et al., 1998; Boevink et al., 1998; Ueda et al., 2001). In this work, we used transient P35S-driven expression plasmids and TMV-based vectors (with or without a functional MP) to investigate localization of GFP- and mRFP-tagged fusion proteins in epidermal cells. Throughout these experiments, the same pattern of intracellular fluorescent localization was observed regardless of the expression system or whether the proteins were tagged with mRFP or GFP. The same pattern of expression was seen in the PMTV hosts N. benthamiana and N. tabacum and the nonhosts Arabidopsis and A. cepa. Moreover, when expressed in PMTV-infected cells, in the presence of viral RNA and virus-encoded proteins, there was no change in the distribution of mRFP-TGB2, whereas mRFP-TGB3 was recruited to vesicles by the virus-expressed TGB2. Thus, although the TGB fusion proteins were not expressed from a PMTV-based vector, we believe that the subcellular localizations of the TGB fusion proteins are representative of their native distribution.
The TGB2/3 Complex Traffics on the Actin-ER Network
The results described in this article show a close association of TGB2 and TGB3 fusion proteins with the ER and perinuclear membranes, and the fusion proteins were colocalized when expressed in the same cell. There was a consistently early association of TGB fusion proteins with the ER network followed by the appearance of motile granular structures. The small motile bodies induced by both TGB fusion proteins (and in which they were colocalized) moved quickly along the ER network and through cytoplasmic strands. We think these motile granules may transport viral RNP complex to the plasmodesmata. No association of TGB fusion proteins was seen with microtubules when the former were introduced into microtubule-labeled plants. Furthermore, oryzalin (which completely fragmented the microtubules as described in Gillespie et al., 2002) had no effect on the subcellular distribution of the TGB fusion proteins. By contrast, movement of the small granules was aligned with actin filaments when expressed in GFP-hTalin plants (where GFP is fused to the F-actin binding domain of human talin; Takemoto et al., 2003), and such movement was completely inhibited by treatment of cells with latrunculin, which disrupted the F-actin–labeled filaments. In higher plants, the cortical ER network is intricately linked to underlying actin cables that function as tracks for Golgi movements (Boevink et al., 1998). It remains possible that the TGB fusion proteins could pass through the trans-Golgi network en route to plasmodesmata. However, BFA had no influence on the localization or movement of the labeled proteins, and the TGB-containing bodies did not colocalize with the Golgi when the latter were labeled with a Golgi-GFP marker (see Boevink et al., 1998). We suggest, therefore, that the TGB2/3 complex uses the actin-ER network to facilitate movement to plasmodesmata. A similar conclusion was reached for the movement of PSLV TGB3 based on colocalization with ER markers at the cell periphery (Zamyatnin et al., 2002; Gorshkova et al., 2003).
Our results show that the TGB3 fusion protein contains the signal for plasmodesmatal targeting. However, colocalization of this protein with the TMV MP (expressed transgenically as a MP-GFP fusion; Roberts et al., 2001) occurred a few cell boundaries behind the leading edge of infection, and mRFP-TGB3 only targeted a subpopulation of those plasmodesmata labeled with TMV MP. Plasmolysis experiments also indicated that TGB3 did not enter the plasmodesmal pore when transiently expressed in onion cells. Both TGB3 and TGB2 fusion proteins remained associated with the plasma membrane and did not form strong links with the cell wall, a finding similar to that reported by Gorshkova et al. (2003) for PSLV TGB3. Therefore, we suggest that at the leading edge of an infection, TGB3 targets the neck region of plasmodesmata but does not enter the plasmodesmal pore. Unlike mRFP-TGB3, TGB2 fusion proteins did not localize to plasmodesmata and mRFP-TGB3 did not appear to label the punctae thought to represent plasmodesmata when coexpressed with GFP-TGB2, although both fusion proteins were clearly colocalized in larger bodies at the periphery of the cell. Colocalization of TGB2 and TGB3 to peripheral bodies at the plasma membrane has been observed when PSLV TGB2 and TGB3 were coexpressed in the same cell (Solovyev et al., 2000; Zamyatnin et al., 2002). This finding may reflect a regulatory role for TGB2. The arrangement of the genetic module provides a mechanism for regulation where (in the systems studied) TGB2 and TGB3 are expressed from the same subgenomic RNA, with TGB3 probably translated by a leaky scanning mechanism (reviewed in Morozov and Solovyev, 2003). Also, the finding that cell-to-cell movement of PVX TGB1 is inhibited in the presence of transgenically expressed TGB2 and TGB3 provides some evidence for a regulated movement process (Yang et al., 2000).
Although TGB2 does not apparently contain a plasmodesmal targeting signal, both TGB2 and TGB3 fusion proteins were able to increase the SEL of plasmodesmata. However, unlike the PVX proteins, no movement of the PMTV proteins to adjacent cells was detected in the permissive host N. benthamiana (Krishnamurthy et al., 2003). Neither did they move out of PMTV-infected cells. Recently, it was shown that PVX TGB2 could interact with TIP, a host factor that interacts with β-1,3 glucanase, which may serve to regulate SEL of plasmodesmata by callose degradation (Fridborg et al., 2003).
Mutation in the Conserved YQDLN Motif Affects Localization of TGB3
The sequence of TGB3 contains a conserved YQDLN motif in the cytoplasmic domain. This sequence may contain a putative Tyr-based sorting signal based on the motif YXXΦ (where X is any amino acid residue, and Φ represents an amino acid with bulky hydrophobic side chain; Marks et al., 1997). Such motifs are found in the cytoplasmic domains, usually the cytoplasmic tails of transmembrane proteins and are involved in internalization and targeting of proteins to subcellular compartments, such as endosomes, lysosomes, or other organelles (Marks et al., 1997; Bonifacino and Dell'Angelica, 1999). This motif is found in several animal viruses that enter cells by endocytosis (Sieczkarski and Whittaker, 2002), and mutagenesis studies on Varicella-zoster virus glycoprotein B showed that two such Tyr-based motifs in the cytoplasmic tail operated to internalize and to traffic the protein to the Golgi (Heineman and Hall, 2001). Recently, Tyr-based motifs have been noted to occur in the MP of Grapevine fanleaf virus and in the MPs of several nepoviruses (Laporte et al., 2003). In this study, the mutation YQDLN to GQDGN caused a change in subcellular localization of TGB3, eliminating the ER localization and the presence of small motile granules. Disruption of this motif in PSLV TGB3 by insertion of four amino acids (RSTD) between the D and L residues abolished localization to the cell periphery (Solovyev et al., 2000). In our experiments, the mutation also affected the targeting to plasmodesmata and efficiency of plasmodesmal gating. The role that this motif plays in the formation and/or sorting of compartments containing the viral MP complex is the subject of current research in our laboratory.
A Role for the Endocytic Pathway in Retrieving TGB Proteins?
FM4-64 is widely used as a marker for the endocytic pathway (Vida and Emr, 1995; Ueda et al., 2001; Grebe et al., 2003; Bolte et al., 2004). It is a lipophilic stain that fluoresces when inserted into the outer leaflet of lipid bilayers, entering cells by endocytosis from the plasma membrane and subsequently labeling the membranes of many intermediate compartments and small organelles, including Golgi stacks and prevacuolar compartments and finally the vacuolar membrane (Bolte et al., 2004). Staining of the TGB2-containing vesicles with FM4-64 revealed that they were derived, at least in part, from components of the plasma membrane and indicates that they are components of the endocytic pathway. The TGB2 fusion protein was seen in these motile vesicles later in the expression cycle, and when coexpressed in the same cell, TGB2 and TGB3 fusion proteins became colocalized to the same structures. Significantly, GFP- or mRFP-TGB3 were never seen to associate with vesicle-like structures when expressed independently and continued to accumulate at the cell periphery. Thus, one of the functions of TGB2 may be to remove excess TGB3 once the former has successfully located the viral RNP to the plasmodesmal pore.
Colocalization of mRFP-TGB2 with GFP-Ara7, a Rab5 ortholog from Arabidopsis, is another indicator that these vesicles represent a component of a plant internal recycling compartment, possibly the endosome (Ueda et al., 2001). Rab5 GTPase is a key regulator of homotypic vesicle fusion in early endosomes (Stenmark et al., 1994; Zerial and McBride, 2004) and may be distributed within discrete domains within the endosomal membrane. Zerial and McBride (2004) have suggested that the Rab5 domain may be the gateway into the early endosome. In plants, Ueda et al. (2001) have suggested that Ara7 locates to the early/late endosome and regulates membrane fusion in the early endocytic pathway.
We also found that TGB2 is needed for TGB3 to enter the endocytic pathway. Both proteins localized to discrete hot spots of fluorescence on vesicles, indicating that they may bind to functionally distinct domains of the same vesicle. This dot and ring structure is very similar to that reported for Ara6 and Ara7 by Ueda et al. (2001). The plant endocytic pathway is not fully described, and doubts continue to be expressed as to the true nature of endosomal compartments in plants (Bolte et al., 2004; Geldner, 2004). However, in animal cells, the early endosome is a dynamic organelle that acts as a sorting station for recycling receptors and their ligands (Holstein, 2002), and enveloped RNA viruses exploit the endocytic sorting mechanisms to facilitate budding (Raiborg et al., 2003). Transport along the recycling pathway depends on a functional actin cytoskeleton (Ayscough, 2004). TGB2 and TGB3 were shown previously to self-interact and also to interact with each other in the yeast two-hybrid system (Cowan et al., 2002). Therefore, it is possible that these interactions are important for homotypic vesicle fusion and multimerization of sorting signals that would improve the efficiency of internalization (Arneson and Miller, 1995).
TGB2 Interacts with RME-8, an Endocytic DnaJ Chaperone
The unique binding of TGB2 to a DnaJ-like plant chaperone belonging to the RME-8 family is further evidence that TGB2 may function in endocytic recycling. The RME-8 proteins do not share sequence similarity outside the J domain, they were first identified from Caenorhabditis elegans and are required for both fluid-phase and receptor-mediated endocytosis (Zhang et al., 2001; Chang et al., 2004). RME-8 is ubiquitous in cells and functions in an early trafficking event before delivery of endocytosed cargo to the lysosomal compartment. In C. elegans, RME-8 localizes to the limiting membrane of large endosomes (Zhang et al., 2001). Recently, Chang et al. (2004) showed that, in Drosophila melanogaster, the J-domain of RME-8 interacts with Hsc70 to function as an essential cochaperone to control clathrin-dependent endocytosis. This function has also been demonstrated with the RME-8 protein auxilin (Ungewickell et al., 1995). The vesicular transport model for the function of endosomes predicts that they are stable cellular compartments from which endosomal contents move from one compartment to another by means of small transport vesicles (Gruenberg and Howell, 1989; Gruenberg and Maxfield, 1995). RME-8 was discovered in a genetic screen for mutants defective in endocytosis in C. elegans (Grant and Hirsh, 1999) and is ubiquitous among several cell types. Using an RME-8:GFP reporter under control of the rme-8 promoter, Zhang et al. (2001) found that RME-8 was localized strongly to vesicles ranging in size from 0.5 to 4.0 μm in diameter, which accumulated the fluid-phase endocytosis marker BSA. These vesicles are remarkably similar to those shown here labeled with TGB2 and Ara7 (e.g., Figures 3J to 3M). To date, there are very few known markers for the endocytic pathway in plants (Bolte et al., 2004), and it will be interesting to determine whether other Rab and RME proteins locate to the structures shown here.
Our results show that the TGB fusion proteins are inserted in the membranes of endosomal vesicles, but there are different possible destinations after entry to the endocytic pathway. For example, proteins could be recycled from early endosomes back to the plasma membrane (either directly or via intermediate compartments), or they could be targeted to the vacuolar pathway for degradation in the lytic vacuole (Holstein, 2002; Bolte et al., 2004). It is possible that association of TGB fusion proteins with membranous compartments in the endocytic pathway indicates that they are being targeted to prevacuolar compartments and eventual degradation in lysosomes, possibly as part of a plant defense mechanism to eliminate virus movement, or triggered as a consequence of prolonged protein overexpression. However, we favor the hypothesis that they are being recycled to the plasma membrane for the following reasons: (1) We do not see the TGB fusion proteins associated with multivesicular bodies or structures characteristic of late endosomes (reviewed in Holstein, 2002) nor any labeling of vacuolar membranes; (2) if the proteins were directed to lysosomes, then it is not clear why the fusion protein–labeled vesicles accumulate in clusters around the nuclear envelope; (3) if the proteins were targeted for degradation, we might expect to see similar localizations with the other overexpressed proteins, such as GFP, mRFP, GFP-sporamin, or GFP-TGB3 or mRFP-TGB3 expressed independently, which does not occur.
The association of PMTV TGB proteins with the endocytic pathway is unique and has not been reported in similar published studies of other plant–viral TGB proteins. However, a recent report suggests similar vesicles are also produced in PVX GFP-TGB2 transgenic plants (Mitra and Verchot-Lubicz, 2004). It is possible that the endocytic pathway is involved in the natural transmission of PMTV, which is transmitted in nature by the soil-borne plasmodiophorid S. subterranea, an intracellular parasite that infects the cells of potato tubers and roots. PMTV is carried within vector zoospores, but it is not known how the virus is acquired or transmitted. The contents of the primary zoospore are injected directly into host cytoplasm, which then develops into a multinucleate plasmodium (Braselton, 1992, 1995). The plasmodium is separated from the host cell by a single unit membrane, so it is conceivable that PMTV requires an endocytic/exocytic mechanism for vector transmission. If so, a similar mechanism might be predicted to occur in other TGB containing soil-borne viruses, such as Beet soil-borne virus, Beet necrotic yellow vein virus, and Peanut clump virus, which are transmitted by the plasmodiophorids Polymyxa betae and Polymyxa graminis.
Positive strand RNA viruses use host cellular membranes for replication and in several plant viruses; for example, TMV (Más and Beachy, 1999), Cowpea mosaic virus (Carette et al., 2002), Brome mosaic virus (Restrepo-Hartwig and Ahlquist, 1996), and Tobacco etch virus (Schaad et al., 1997) replication is associated with ER derived membranes. TMV and Peanut clump virus replication is thought to occur in close association with perinuclear ER membranes (Más and Beachy, 1999; Dunoyer et al., 2002). Some viruses induce the formation of spherules or invaginations of the membrane that contain the RNA template, replicase, and other replication-associated factors on the cytoplasmic face. The spherules bring together the replication factors to enhance replication and separate them from possible host defense factors. The nascent RNA strands are thought to exit from the neck of the spherules and presumably then interact with components of the movement complex (see reviews in Ahlquist et al., 2003; Noueiry and Ahlquist, 2003; Salonen et al., 2005). The fact that PMTV TGB proteins are associated with ER and perinuclear membranes may indicate that they, too, are located close to sites of positive strand RNA synthesis. The TGB-associated vesicles are not thought to contain replication complexes because they appear to form at the plasma membrane and their membrane is not derived from ER, but we cannot rule out the possibility that PMTV replicates on endosomal membranes and that another function of TGB2 is to help recruit such membranes.
A Model for PMTV TGB Functions
A schematic model to interpret our results is shown in Figure 4M. The model depicts different phases of movement we would hypothesize from our results, namely trafficking to plasmodesmata and recycling of membrane anchored components. TGB2 and TGB3 colocalize in the same small motile granules that move along the actin-ER network and are targeted to the neck of plasmodesmata by TGB3. The granules dock at the neck of the plasmodesmal pore, and the TGB2/3 proteins function in tandem to increase the SEL of the pore. At present, it is unclear why both TGB2 and TGB3 possess the capacity to gate plasmodesmata, although studies of other TGB viruses have reported that both these proteins possess gating capacity (Krishnamurthy et al., 2003; Mitra et al., 2003). At or near the plasmodesmal pore, we hypothesize that there is a transfer of the TGB2/3 complex from the ER to the plasma membrane (see also Oparka, 2004) and delivery of viral RNP complex, which would move through the plasmodesmal pore. After delivery, the TGB2 and TGB3 proteins are recycled together through the endocytic pathway. This model is also consistent with recent data showing that TGB2 and TGB3 are required for localization of GFP-TGB1 to the periphery of the cell and movement into adjacent cells (Zamyatnin et al., 2004).
The Link between Endocytosis and Viral Movement
We believe our findings for TGB2 and TGB3 are relevant to the native distribution of the TGB proteins because they have been validated in the presence of infectious virus. In the future, the recycling model must be tested to determine whether viral MPs are recycled to the cell interior to resume the trafficking of viral RNA complexes (under control of the virus) or whether they are targeted to the vacuole for subsequent degradation by the plant cell (under control of the host). The above model is designed to stimulate such experimentation and to initiate further studies of the relationship between plant–viral MPs and the plant endocytic recycling pathway. To date, entry of viruses into plant cells by the endocytic pathway has not been documented (Hull, 2002), although the mechanism for entry of plant viruses into tissues with no symplastic connection, such as embryos or zoospores, is unknown. However, the use of the endocytic pathway for virus entry is common in the animal kingdom (Sieczkarski and Whittaker, 2002). Our observations that the TGB2/3 complex of PMTV may use the endocytic pathway raises important questions for virus movement in plants and lends further support to the growing body of evidence that viral MPs may interact with components of the plant endomembrane recycling pathway, including Rabs, syntaxins, and DnaJ-based chaperones (reviewed in Oparka, 2004). Such components are likely to be the elusive host factors alluded to in many viral movement models (Oparka, 2004).
METHODS
TMV-Based Clones
TMV clones expressing GFP-TGB2 and GFP-TGB3 were described by Cowan et al. (2002). These clones were used to prepare the TMV-based clones with defective MP. The respective plasmid DNA was digested with ApaI, followed by treatment with T4 DNA polymerase and religation. This procedure causes a frame-shift mutation in the TMV MP, resulting in movement-incompetent clones.
Transient Expression Plasmids
The TGB3 and GFP genes were amplified separately from the TMV-GFP-TGB3 vector template, and the products of each reaction were combined in an overlap extension PCR (Higuchi et al., 1988) to fuse the TGB3 gene to the 3′ terminus of the GFP sequence. The GFP-TGB3 fragment was ligated into plasmid vector pRTL2 (Restrepo et al., 1990), which expresses GFP-TGB3 under the control of the 35S promoter (P35S). Similarly, overlap extension PCR was used to clone the TGB2 and TGB3 genes as fusions to the 3′ terminus of mRFP. The TGB2 and TGB3 genes were amplified from a plasmid containing the PMTV RNA2 sequence (Cowan et al., 2002), and mRFP was amplified from plasmid pmRFP1 (Clontech, Palo Alto, CA). The fusions were ligated into pRTL2 as above.
Two mutant constructs were prepared from the P35S-GFP-TGB3 using the QuikChange kit (Stratagene, La Jolla, CA) where the TGB3 protein sequence -Y89QDLN93 was modified to YAAAA or GQDGN. GFP-Ara7 was produced by PCR amplification of Ara7 cDNA and cloned into pVKH 18-En6 (Batoko et al., 2000) so that GFP is expressed as an N-terminal fusion.
Plant Material
Several transgenic plants were used for colabeling experiments, including the following: Nicotiana tabacum, TMV-30K-GFP (TMV MP-GFP) (Reichel and Beachy, 2000); N. benthamiana, SmRS.GFP-TUA6 (tua-GFP) (Ueda and Hashimoto, 1999); N. benthamiana, HDEL-GFP (erGFP) (Carette et al., 2000); Arabidopsis thaliana, sialyl transferase-GFP (Boevink et al., 1998); A. thaliana, GFP-hTalin (Takemoto et al., 2003). Plants were grown from seed and maintained at 28°C in a Snijder climatic cabinet (Snijder, Tilburg, Holland) with a photoperiod of 16 h. Light intensity was ∼400 μE m−2 s−1. Plants were used for experiments when they were between 15 and 60 d old.
Microscopic Imaging of Intact Leaves
Intact leaves, infected with the different vectors expressing GFP- or mRFP-TGB2 and TGB3 fusions, were fixed to microscope slides using double-sided adhesive tape (Sellotape GB, Dunstable, UK). Confocal images were obtained using a Leica TCS SP spectral confocal laser-scanning microscope (CLSM; Leica Microsystems, Heidelberg, Germany) with Leica water-dipping lenses. GFP was excited at 488 nm and emissions collected at 500 to 530 nm. mRFP was excited at 568 nm and emissions collected at 580 to 600nm. For dual imaging of fusions with both fluorescent proteins, images were taken sequentially to avoid bleed-through of signals (Leica TSC Confocal Systems user manual). Post-acquisition processing of images was done using Adobe Photoshop 7.0 software (Mountain View, CA).
Preparation of Infectious Transcripts and Plant Inoculation
Run-off transcripts were synthesized from viral vectors using a T7 transcription kit (Ambion, Austin, TX). For manual inoculation, transcripts were rubbed onto aluminium oxide-dusted N. benthamiana leaves. Two and 4 d postinoculation, inoculated leaves were detached from plants and examined under the CLSM. At this time, infection sites containing the virus vector with a functional MP covered an area of between 20 and 100 epidermal cells.
Microprojectile Bombardment
Plasmid DNAs and transcripts from viral constructs were introduced into source leaf epidermal cells by particle bombardment using a Handgun essentially as described by Gal-On et al. (1997). Briefly, ∼2 μg of plasmid DNA or RNA transcript was mixed with 5 μL of ethanol and 0.55 mg of 1-μm gold micro carrier (Bio-Rad Laboratories, Hercules, CA) in ethanol to give a final volume of 18 μL. The nucleic acid-gold mixture (2 μL) was applied to a discharge assembly (13-mm Plastic Swinney Filter Holder; PALL Gelman Laboratory, Ann Arbor, MI) and ethanol allowed to evaporate before bombardment. Bombarded leaves were observed under the CLSM after 24 to 48 h. In cobombardment experiments, equal amounts of the plasmids and RNA transcript (2 μg in total) were mixed before addition of the gold particles.
PMTV-Infected Plants
For the experiments with virus-infected leaves, PMTV isolate T (Cowan et al., 2002) was inoculated to aluminium oxide-dusted N. benthamiana leaves and plasmid DNA bombarded, as described above, to the same leaves 24 to 72 h later. CLSM observations were made 2 d after bombardment. Virus infection was confirmed by tissue print immunoassay (Toth et al., 1999) or RT-PCR on extracts from imaged leaves using primers to amplify CP or TGB1 as described by Torrance et al. (1999).
Plasmodesmatal Gating Experiments
Source leaf epidermal cells (N. benthamiana and N. tabacum) were cobombarded with mixtures of plasmid DNA. The mixtures comprised P35S-GFP-sporamin (47 kD; Oparka et al., 1999) and either P35S-mRFP-TGB2, P35S-mRFP-TGB3, P35S-mRFP-TGB3 mutant (YQDL to GQDG), or P35S-mRFP. Cobombarded cells were examined by CLSM at 2 d after bombardment.
Onion Cell Plasmolysis
To allow high-resolution imaging of plasmolysis, plasmid DNAs expressing GFP- or mRFP-TGB2 and TGB3 fusions under the control of the 35S promoter were introduced into onion epidermal cells by particle bombardment using the method described above. The epidermis was bombarded while still attached to the underlying leaf (bulb) tissue and maintained attached to the leaf in humid plastic containers at room temperature. After 24 to 48 h, the epidermis was examined for transformed cells. Tissue showing fluorescent protein expression in transformed cells was soaked in water before carefully removing the epidermis and attaching to a plastic frame to allow easy handling. The tissue was then floated on water for 40 min before being incubated in either water or 750 mM sorbitol for 30 min. Tissue samples were then imaged by CLSM as described above.
Chemical Treatments
Leaves from appropriate transgenic plants were bombarded with plasmid DNA, and at 24 to 48 h after bombardment, these and control leaves untreated were detached and infiltrated with either BFA (100 μg mL−1), latrunculin (1 μg mL−1), oryzalin (20 μg mL−1), or water. Leaf tissue was infiltrated through stomata on the abaxial leaf surface. The infiltrated tissue was then floated with its abaxial surface in the inhibitor solutions and maintained under a light source for the duration of the treatment. After infiltration, the leaves were floated on BFA or latrunculin solutions for 1 to 2 h and oryzalin for 2 h before imaging as described above.
FM4-64 Staining
Samples of leaf tissue, either untreated or 24 to 48 h after bombardment with plasmid DNA, were infiltrated through stomata on the abaxial leaf surface with a solution of FM4-64 (Molecular Probes, Eugene, OR) at a concentration of 50 μM in distilled water. Samples were kept under lights in a sealed humid container at room temperature and examined by CSLM as described above after 1 to 18 h.
TGB2 Overlay Blots
A λ-ZAP cDNA library from tobacco leaf N. tabacum var SR1 (Stratagene, La Jolla, CA) was excised from the λ-ZAP II vector according to the manufacturer's instructions. The plasmid library was plated on LB medium containing ampicillin. Expression was induced with isopropylthio-β-galactoside. Nitrocellulose filters were placed on the replica plates, and after blotting the filters were processed as described in the Stratagene picoBlue immunoscreening kit instruction manual. Briefly, the filters were overlaid with purified preparations of Escherichia coli–expressed TGB2 (prepared as described in Cowan et al., 2002) at 0.5 μg/mL and incubated for 16 h at 4°C. Filters were washed, blocked with 1% BSA in Tris-buffered saline, and incubated with anti-TGB2 antibody (1/200) and anti-rabbit AP conjugate (1/10,000). Then, filters were washed and incubated with substrate solution (BCIP/NBT liquid substrate system; Sigma-Aldrich, St. Louis, MO).
Sequencing and Sequence Alignments
DNA sequencing was done using the BigDye Terminator cycle sequencing kit (Applied Biosystems, Foster City, CA). Relevant amino acid sequences were acquired from public databases, and sequence alignments were done using Vector NTI software (Informax; Invitrogen, Carlsbad, CA). BLAST programs (Altschul et al., 1997) were used to search for sequence similarities in databases.
Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession number AJ704986 (partial sequence of Ntp1).
Acknowledgments
We thank Mark Phillips (Scottish Crop Research Institute) for help with the statistical analysis and Amanda Kotzer and Chris Hawes (Oxford Brookes University) for providing the GFP-Ara7 construct. Scottish Crop Research Institute is grant aided by the Scottish Executive Environment and Rural Affairs Department.
Footnotes
-
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: Lesley Torrance (ltorra{at}scri.sari.ac.uk).
-
Article, publication date, and citation information can be found at www.plantcell.org/cgi/doi/10.1105/tpc.104.027821.
-
↵ w⃞ Online version contains Web-only data.
- Received September 20, 2004.
- Accepted November 5, 2004.
- Published December 17, 2004.