Inhibition of tobacco mosaic virus movement by expression of an actin-binding protein.

The tobacco mosaic virus (TMV) movement protein (MP) required for the cell-to-cell spread of viral RNA interacts with the endoplasmic reticulum (ER) as well as with the cytoskeleton during infection. Whereas associations of MP with ER and microtubules have been intensely investigated, research on the role of actin has been rather scarce. We demonstrate that Nicotiana benthamiana plants transgenic for the actin-binding domain 2 of Arabidopsis (Arabidopsis thaliana) fimbrin (AtFIM1) fused to green fluorescent protein (ABD2:GFP) exhibit a dynamic ABD2:GFP-labeled actin cytoskeleton and myosin-dependent Golgi trafficking. These plants also support the movement of TMV. In contrast, both myosin-dependent Golgi trafficking and TMV movement are dominantly inhibited when ABD2:GFP is expressed transiently. Inhibition is mediated through binding of ABD2:GFP to actin filaments, since TMV movement is restored upon disruption of the ABD2:GFP-labeled actin network with latrunculin B. Latrunculin B shows no significant effect on the spread of TMV infection in either wild-type plants or ABD2:GFP transgenic plants under our treatment conditions. We did not observe any binding of MP along the length of actin filaments. Collectively, these observations demonstrate that TMV movement does not require an intact actomyosin system. Nevertheless, actin-binding proteins appear to have the potential to exert control over TMV movement through the inhibition of myosin-associated protein trafficking along the ER membrane.

2002; Heinlein and Epel, 2004;Lucas and Lee, 2004). The most extensively studied MP is that of Tobacco mosaic virus (TMV). This protein localizes to PD and modifies the size-exclusion limit of the pores (Tomenius et al., 1987;Wolf et al., 1989;Atkins et al., 1991;Ding et al., 1992a;Moore et al., 1992;Oparka et al., 1997). The MP also has the capacity to bind single-stranded nucleic acids in vitro (Citovsky et al., 1990), and since complexes of MP and viral RNA (vRNA) were isolated from TMV-infected plants (Dorokhov et al., 1983(Dorokhov et al., , 1984, vRNA movement likely occurs in the form of a ribonucleoprotein complex. Indeed, the viral capsid protein is dispensable for cell-to-cell movement (Siegel et al., 1978;Dawson et al., 1988;Saito et al., 1990;Hilf and Dawson, 1993), indicating that the virus moves between cells in a nonencapsidated form. During infection, the protein interacts with endoplasmic reticulum (ER) membranes and microtubules, suggesting a role of these cellular components in vRNA movement (Heinlein et al., 1995(Heinlein et al., , 1998. The intercellular transport of vRNA involves the occurrence of mobile, microtubule-proximal, MP-containing particles in cells undergoing the spread of infection (Boyko et al., 2007). Transient expression experiments provided evidence that the MP-containing particles are associated with RNA and that they move in association with ER membranes, thereby continuously undergoing transient interactions with microtubules . The movement of the particles involves microtubule polymerization activity , consistent with the ability of MP to interact with GFP-tagged microtubule end-binding protein 1 (AtEB1a; Brandner et al., 2008) and g-tubulin .
The observed ER-mediated transport of MPcontaining particles is consistent with the general perception that TMV movement involves the ER network. ER membranes are continuous between cells through PD (Ding et al., 1992b;Heinlein and Epel, 2004) and represent a potential pathway for the movement of membrane-associated replication complexes from the infected cells into adjacent cells. This model of TMV movement is supported by observations indicating that ER membrane proteins can laterally diffuse within the membrane (Runions et al., 2006;Guenoune-Gelbart et al., 2008) and move from cell to cell (Guenoune-Gelbart et al., 2008).
Since the ER is tightly associated with the actin network (Boevink et al., 1998), microfilaments could play an important role in supporting the ER-mediated targeting of MP and/or vRNA to PD. Consistent with this hypothesis, the treatment of plants with actin polymerization inhibitors reduces both MP particle trafficking  and the efficiency by which MP accumulates in PD (Wright et al., 2007). Moreover, inhibition of the actin cytoskeleton for several days by either actin silencing (Liu et al., 2005) or inhibitor treatment (Kawakami et al., 2004) reduced the efficiency of TMV movement. The role of actin in supporting the spread of infection may be indirect, however. Although an interaction of MP with actin filaments has been reported (McLean et al., 1995), actin depolymerization reduces but does not fully inhibit PD targeting of MP (Wright et al., 2007). In addition, when actin inhibitor was present for longer than 18 h, no effect on the accumulation of MP in PD was observed (Prokhnevsky et al., 2005). These findings are consistent with the proposal that the targeting of MP to PD occurs by diffusion in the ER membrane (Wright et al., 2007) and that the actin cytoskeleton is not required for this mechanism.
To further clarify the role of the actin cytoskeleton in TMV movement, we applied latrunculin B (LatB) treatments to cells in which actin filaments were labeled by expression of the actin-binding domain 2 (ABD2) of Arabidopsis fimbrin (AtFIM1) fused to GFP (ABD2: GFP; Sheahan et al., 2004). Our observations demonstrate that TMV cell-to-cell movement can continue in the absence of an intact actin cytoskeleton and indicate that TMV movement is primarily ER membrane mediated. Nonetheless, ER-associated actin filaments and actin-binding proteins may play an important role in controlling TMV movement efficiency.

Labeling of Actin Filaments with ABD:GFP
To visualize actin filaments in vivo, we expressed ABD2:GFP from a binary vector under the control of the cauliflower mosaic virus 35S promoter. Transient expression of ABD2:GFP in tobacco (Nicotiana tabacum) BY-2 cells resulted in the green fluorescent labeling of a dense actin network (Fig. 1, A and B). The effective labeling of actin filaments with ABD2:GFP was confirmed by costaining the cells with rhodaminephalloidin ( Fig. 1, C-E). Actin filaments were also effectively visualized in agroinfiltrated Nicotiana benthamiana leaves, where the coexpression of ABD2:GFP with ABD2 fused to red fluorescent protein (ABD2: RFP) resulted in exactly coinciding patterns of greenand red-labeled filaments (Fig. 1, F-H). The pattern of actin filaments in agroinfiltrated leaf tissues was similar to the pattern detected in ABD2:GFP transgenic plants (Fig. 1, I-M); however, the fluorescence emitted from ABD2:GFP in the agrotransfected tissues increased over time. Immunoblot analysis (Fig. 1N) confirmed that the average steady-state levels of transiently expressed ABD2:GFP and derived products in agroinfiltrated tissues started to exceed the level of ABD2:GFP expressed in transgenic plants at about 3 d postagroinfiltration (3 dpa; lane 6). Consistent with relatively low ABD2:GFP expression levels, the ABD2: GFP transgenic plants developed normally and were indistinguishable from nontransgenic wild-type plants (Fig. 1O). To avoid potential adverse effects produced by accumulating ABD2:GFP in agroinfiltrated plant tissues, subsequent experiments involving transient expression of the protein were performed at 1.5 dpa (if not otherwise noted).

Transient, But Not Transgenic, ABD2:GFP Expression Interferes with Myosin-Based Motility
The actin cytoskeleton is dynamic and undergoes continuous rearrangements (Staiger and Blanchoin, 2006). We used time-lapse microscopy to examine the effect of ABD2:GFP expression on actin filament dynamics in leaf epidermal cells of agroinfiltrated (at 1.5 dpi) or transgenic plants. Whereas the ABD2:GFPlabeled actin cytoskeleton in ABD2:GFP transgenic plants ( Fig. 2A) showed dynamic movements of individual filaments (Supplemental Movie S1), the ABD2: GFP-labeled actin cytoskeleton in agroinfiltrated ABD2:GFP-expressing tissues (Fig. 2B) was static (Supplemental Movie S2). Thus, transient expression of ABD2:GFP causes stabilization of an otherwise dynamic actin cytoskeleton.
Since a functional actin-myosin system mediates ER and Golgi motility in plants (Boevink et al., 1998;Lichtscheidl and Baluska, 2000), we investigated the effect of ABD2 expression on ER and Golgi structure and motility by coexpression of specific, fluorescent protein-tagged markers. The structure and dynamic behavior of the ER is conveniently revealed by confocal microscopy of epidermal cells of N. benthamiana line 16c, which is transgenic for GFP carrying an HDEL ER retention sequence (Ruiz et al., 1998). The ER network in these plants consists of the typical polygonal pattern of membrane sheets and tubules ( Fig. 2C) and is highly dynamic. A dynamic ER network was also revealed by time-lapse analysis of wildtype plants in which the ER was labeled by transient expression of GFP:HDEL, RFP:HDEL, or TM17:GFP, an ER transmembrane peptide fused to GFP (Brandizzi et al., 2002;data not shown). Transient expression of RFP:HDEL in ABD2:GFP transgenic plants revealed that transgenic ABD2:GFP expression does not interfere with the structure and dynamic movements of the ER (Fig. 2, D and E; Supplemental Movie S3). However, upon transient expression of ABD2:RFP in epidermal cells of 16c plants, varying effects on ER structure were observed at 1.5 dpa. Some cells exhibited no obvious defects in the structure of the ER network (Fig. 2, F and G), whereas other cells showed clear defects such as gaps in the ER network (Fig. 2, H and I), the fusion of ER membranes into large sheets (Fig. 2, J and K), or the partial (Fig. 2L) to almost complete ( Fig. 2M) contraction of the ER to the vertices. Transient ABD2:GFP expression also caused a partial or complete inhibition of ER membrane movements; this was seen irrespective of whether the ER structure was altered ( The motility of Golgi stacks depends on specific class XI myosins (Avisar et al., 2008b;Peremyslov et al., 2008;Prokhnevsky et al., 2008;Sparkes et al., 2008), requires an intact and dynamic actin cytoskeleton (Nebenfü hr et al., 1999;Brandizzi et al., 2002), and appears to occur with the membranes of the ER (Runions et al., 2006). To investigate whether expression of ABD2:GFP affects actin-myosin-based Golgi motility, we monitored the motility of Golgi stacks upon transient expression of the red fluorescent, tdTomato-tagged, cis-Golgi marker GmMan1 (GmMan1: tdTomato;Nebenfü hr et al., 1999;Fig. 3, A and B). Projections of pixel intensities between frames in timelapse movies display the "paths" of translocating Golgi stacks through the cell (Fig. 3, C and D). In time-lapse movies taken for the duration of 20 s, the velocity of observed Golgi stack movements varied between 0 and 4.1 mm s 21 . These values are within the range of velocities reported for Golgi stacks in differ-   (Boevink et al., 1998;Nebenfü hr et al., 1999;Avisar et al., 2008b;Prokhnevsky et al., 2008;Sparkes et al., 2008). Thus, Golgi trafficking appears not to be affected by transient expression of Golgi marker GmMan1:tdTomato. Expression of this marker in transgenic ABD2:GFP plants revealed dynamic Golgi trafficking comparable to that in wild-type plants ( To investigate whether ABD2:GFP expression may also affect microtubule dynamics, we expressed the Arabidopsis AtEB1a fused to GFP (AtEB1a:GFP; Chan et al., 2003). Expression of AtEB1a:GFP has been reported to reveal the dynamic behavior of microtubules upon transgenic or transient expression in Arabidopsis, tobacco, and N. benthamiana cells (Chan et al., 2003;Mathur et al., 2003;Van Damme et al., 2004;Dixit et al., 2006;Brandner et al., 2008). However, transient coexpression of ABD2:GFP had no significant effect on microtubule polymerization (data not shown), indicating that microtubules are not affected by the inhibition of membrane movements induced by transient ABD2:GFP expression.

TMV Movement in the Presence of an Actin-Disrupting Agent
To investigate the role of actin filaments in TMV movement, we used TMV-MP:RFP, a TMV derivative that encodes full-length MP in fusion to RFP under the control of the MP subgenomic promoter . Following inoculation of N. benthamiana plants, leaf tissues carrying infection sites were infiltrated at 3.5 d postinfection (dpi) with buffer only or with buffer containing 10 mM or 100 mM LatB. Images of individual infection sites were acquired at 3.5 and 4.5 dpi (Fig.  4B), and the fold increase in size of each site over the intervening 24 h was determined. The fold increase reflects the extent of cell-to-cell progression of TMV infection within the assay period. Multiple infection sites were measured in three independent experiments. As shown in Figure 4, A and B, TMV-MP:RFP infection sites in leaves of wild-type plants as well as in leaves of ABD2:GFP transgenic plants continued to expand efficiently irrespective of the presence of LatB. Microscopic analysis of infection sites in leaves of ABD2:GFP transgenic plants revealed that the actin cytoskeleton in mock-treated tissues was intact (Fig.  4C), whereas the actin network was disrupted in the cells of leaves treated with LatB (Fig. 4D). Furthermore, after 24 h of LatB treatment, the structure of the ER network was also affected. Consistent with a previous report (Wright et al., 2007), we observed that the ER was contracted to the vertices (Fig. 4F).
TMV Movement in the Presence of Transiently Expressed ABD2:GFP As described above, transient expression of ABD2: GFP leads to inhibition of the dynamic myosin-depen-dent movements of Golgi stacks without affecting the integrity of the actin filaments. To test whether the apparent inhibition of myosin-dependent transport activity has an effect on the efficiency of TMV movement, leaves carrying TMV-MP:RFP infection sites were agroinfiltrated at 3 dpi. After 1.5 dpa, the expression of ABD2:GFP was verified by fluorescence microscopy and images of individual infection sites were acquired. Subsequently, the leaf samples were either mock treated or treated with LatB as described for wild-type and ABD2:GFP transgenic plants, and the individual infection sites were again photographed 24 h later (thus at 2.5 dpa). As a control for the agroinfiltration experiment, the inoculated leaves of control plants were not infiltrated, infiltrated with water, infiltrated with agrobacteria carrying an empty binary vector, or infiltrated with agrobacteria containing a binary vector that encodes free GFP. As summarized in Figure 5A, TMV-MP:RFP infection sites expanded efficiently in these control leaves. However, the spread of infection was significantly inhibited in leaves transiently expressing ABD2:GFP. Importantly, the spread of infection in the ABD2:GFP-expressing tissue was not inhibited if the ABD2:GFP-expressing leaves were treated with LatB, indicating that tran-  siently expressed ABD2:GFP interferes with the spread of TMV-MP:RFP infection through a mechanism requiring actin filaments. Microscopic analysis of ABD2:GFP revealed that the actin cytoskeleton in mock-treated tissues was intact (Fig. 5B), whereas it was disrupted in LatB-treated tissues (Fig. 5,C and D). Consistent with the LatB treatment results obtained with wild-type and ABD2:GFP transgenic plants, LatB had no effect on TMV-MP:RFP movement in control leaves that were not infiltrated or were infiltrated either with water or with agrobacteria carrying either an empty plasmid or a GFP-encoding binary plasmid. We also performed a reverse experiment in which the leaves were first agroinfiltrated and then infected with virus 1.5 d later. As displayed in Table I In TMV-MP:RFP infection sites, the subcellular localization patterns of MP:RFP are similar to those described for the MP:GFP of TMV-MP:GFP (Heinlein et al., 1998;Supplemental Fig. S1) and include the association with microtubules (Heinlein et al., 1998;Boyko et al., 2000b;Ashby et al., 2006). Because immunostaining procedures applied to fixed protoplasts have indicated that MP may also have the capacity to align along phalloidin-rhodamine-labeled microfilaments (McLean et al., 1995), we investigated whether some of the in vivo MP:RFP-decorated filaments may represent actin filaments. However, confo-  cal microscopy of infection sites in ABD2:GFP transgenic plants revealed that MP:RFP-associated filaments are clearly distinct from the ABD2:GFP-labeled microfilaments (Fig. 6, A-F). Nevertheless, some MP:RFPassociated aggregate structures (ER-associated viral replication sites) occurred in the vicinity of ABD2: GFP-labeled filaments (Fig. 6, A-C and G-I), which is consistent with the actin-dependent development and trafficking of these structures (Heinlein et al., 1998;Kawakami et al., 2004;Liu et al., 2005).

DISCUSSION
In this study, we investigated the role of the actin cytoskeleton as well as the potential involvement of interactions between MP and actin filaments in TMV movement. To address the in vivo localization of MP with respect to actin filaments and to verify the effect of treating plant tissue with the actin polymerization inhibitor LatB, we labeled the actin filaments by expression of the actin-binding protein ABD2:GFP. In-terestingly, we found that transient, unlike stable, expression of this marker interferes with actomyosindependent motility, as indicated by the lack of Golgi stack movements. Thus, in our analysis of the role of the actin cytoskeleton in TMV movement, ABD2:GFP had a dual role: in addition to serving as a marker to visualize the actin network, its transient expression enabled us to apply this protein as an inhibitor of myosin-based motility in the presence of an intact actin cytoskeleton. Our results demonstrate that TMV movement is inhibited upon inhibition of myosinbased motility by transient expression of ABD2:GFP. This inhibition is dependent on the actin cytoskeleton and eliminated upon disruption by LatB. Thus, whereas the inhibition of TMV movement with transiently expressed ABD2:GFP is actin dependent, the mechanism that supports TMV movement is actin independent. Collectively, these findings are consistent with the proposal that the intracellular transport of TMV occurs via the ER and that the adjacent actin cytoskeleton may exert control over ER-mediated transport via myosin bridges.

Inhibition of TMV Movement by ABD2:GFP through Interference with Macromolecular Trafficking in the ER Membrane
The observation that transient ABD2:GFP expression interfered with myosin-mediated motility is in agreement with several other reports indicating that the expression of GFP fusions with actin-binding proteins can interfere with actin dynamics and organization, even though they are valuable tools for evaluating cytoskeletal functions (Kost et al., 1998;Mathur et al., 1999;Fu et al., 2001;Brandizzi et al., 2002;Ilgenfritz et al., 2003;Holweg et al., 2004;Ketelaar et al., 2004;Weerasinghe et al., 2005;Xu and Scheres, 2005;Holweg, 2007;Wang et al., 2008). The inhibition of myosin-based dynamics upon transient expression of ABD2:GFP in our N. benthamiana system is probably related to the ABD2:GFP expression level, which during the time of the experiment (1.5 and 2.5 dpa) was in excess of the level of ABD2:GFP in transgenic plants. The mechanism that inhibits myosinmediated motility and TMV movement upon transient ABD2:GFP expression is LatB sensitive and, therefore, actin dependent. Based on several reports describing the inhibition of kinesin-based movements by microtubule-associated proteins (Lopez and Sheetz, 1993;Ebneth et al., 1998;Ashby et al., 2006;Dixit et al., 2008;Vershinin et al., 2008), it seems probable that excess binding of ABD2:GFP to the actin cytoskeleton interferes with the translocation of myosin motor proteins along the filaments. The observed restriction of Golgi and TMV trafficking likely involves the inhibition of class XI myosins, which move along actin filaments (Tominaga et al., 2003) and are responsible for the dynamic movements of Golgi complexes and other organelles (Avisar et al., 2008b;Peremyslov et al., 2008;Prokhnevsky et al., 2008;Sparkes et al., 2008) as well as for those of the ER (Kashiyama et al., 2000;Morimatsu et al., 2000;Kimura et al., 2003;Yokota et al., 2009). The myosin-driven Golgi movements occur along actin filaments that align the ER (Boevink et al., 1998) and have been proposed to reflect myosin-supported movements of ER membrane proteins (Runions et al., 2006). The inhibition of Golgi trafficking upon transient ABD2:GFP expression may thus reflect inhibition of the trafficking of protein complexes in the ER membrane. Given that TMV movement occurs via diffusion in the ER membrane (Brill et al., 2000;Wright et al., 2007;Guenoune-Gelbart et al., 2008;Sambade et al., 2008), we propose that the ABD2:GFP-mediated restriction in TMV movement occurs through obstruction of the ER membrane by immobilized protein complexes (Fig. 7). We speculate that the inhibitory mechanism involves interference with class XI myosins; nevertheless, class VIII myosins may also have a role, since they are associated with PD (Reichelt et al., 1999;Volkmann et al., 2003;Golomb et al., 2008). Although specific inhibition of class VIII myosin cargo binding was reported to have no effect on the targeting of the TMV MP to PD (Avisar et al., 2008a), inhibiting the translocation of this myosin class along actin filaments could still lead to hindrance in the cellular pathway that supports the spread of infection. We also consider the possibility that ABD2:GFP expression may trigger inhibition of TMV movement already during the formation or functioning of ER-associated TMV replication complexes. Nonetheless, a significant influence of transient ABD2:GFP expression on virus replication efficiency appears unlikely, since the fluorescence intensity of TMV-MP:RFP infection sites was not decreased irrespective of the presence of this protein.

TMV Movement Is Independent of Actin
The ability of LatB to eliminate the inhibition of TMV movement caused by transiently expressed ABD2:GFP as well as the absence of any significant effect of LatB on TMV movement efficiency in wildtype or ABD2:GFP transgenic plants demonstrate that an intact actin cytoskeleton is dispensable for TMV movement. Since MP traffics in the ER membrane (Brill et al., 2000;Wright et al., 2007;Guenoune-Gelbart et al., 2008;Sambade et al., 2008), this finding is in agreement with the reported ability of ER membrane proteins to continue their trafficking in the membrane upon disruption of actin filaments with LatB (Runions et al., 2006). Nevertheless, although an effect of LatB treatment on the spread of infection was not observed, the absence of intact actin filaments has important effects with respect to the efficiency of PD targeting. Thus, disruption of the actin cytoskeleton slowed down the trafficking of ER membraneassociated MP  and of other ER proteins (Runions et al., 2006). LatB treatment also reduced the accumulation of MP in PD (Wright et al., 2007). These effects of LatB on ER-mediated transport have the potential to reduce the number of virus genomes that spread between cells over time. However, because the number of transported genomes is usually not a limiting factor in the spread of infection (Sacristan et al., 2003;Li and Roossinck, 2004), the effects of LatB do not necessarily translate into an obvious effect at the level of infection sites. Thus, even though we did not observe an effect of LatB treatment on the spread of infection in N. benthamiana leaf tissues under our conditions, we do not exclude the possibility that a potential reduction in the number of transported virus genomes may represent a critical factor for infection in other hosts or tissues. Moreover, an inhibition of virus movement in LatBtreated cells may depend on the experimental conditions. As concluded here, TMV movement occurs undisturbed upon disruption of the actin cytoskeleton for 24 h; however, according to previous reports, TMV movement can be inhibited if the disruption of the actin cytoskeleton continues for several days (Kawakami et al., 2004;Liu et al., 2005). We note that our conclusions are based on the analysis of over 150 individual infection sites treated with LatB (Figs. 4 and 5), whereas the previous reports were based on smaller sample sizes. Nevertheless, inhibition of TMV movement following prolonged LatB treatment may be expected, since the absence of an intact actin network gradually induces the contraction of the ER (Wright et al., 2007). The contraction of the ER is not observed if the drug is applied during shorter time periods (Boevink et al., 1998;Sambade et al., 2008). Thus, the general structure and function of the ER is independent of actin and persists over a certain time in the presence of LatB. It might be considered that the alterations in ER structure observed in ABD2: GFP-overexpressing cells contributed to the ABD2: GFP-induced inhibition of TMV movement. However, this possibility appears unlikely, since the inhibition by ABD2:GFP was abolished upon LatB treatment; rather, LatB should enhance inhibition if mediated by an aberrant ER structure. We also observed the contraction of ER following the application of LatB for 24 h, thus at the end of the time window during which our TMV movement assay was performed and TMV efficiently moved from cell to cell. Since LatB-induced ER contraction occurs gradually, we assume that the changes in ER structure did not reach inhibitory levels within the 24-h observation period of our assay. However, longer LatB treatments may affect the structure of the ER to an extent that interferes with TMV movement.

MP Does Not Align to Actin Filaments
Neither in cells infected with TMV-MP:RFP nor in cells ectopically expressing MP:RFP did we observe any alignment of MP:RFP to ABD2:GFP-labeled actin filaments. Although MP:RFP-associated filaments were observed, these filaments were not labeled with ABD2:GFP. We cannot exclude the possibility that ABD2:GFP binding to actin filaments interferes with MP:RFP binding. However, since other studies have demonstrated that the filaments to which MP binds in plant and mammalian cells are microtubules (Heinlein et al., 1995(Heinlein et al., , 1998Boyko et al., 2000aBoyko et al., , 2000bAshby et al., 2006;Ferralli et al., 2006), this possibility appears unlikely. Our observation is consistent with the finding that MP does not bind to actin filaments in mammalian cells (Ferralli et al., 2006), and actin is highly conserved between plants and mammals. The lack of alignment of MP:RFP to ABD2:GFP-tagged actin filaments in living cells does not confirm the previous observations by McLean et al. (1995), who showed the alignment of antibody-stained MP to phalloidin-rhodamine-labeled filaments in chemically fixed TMV-infected protoplasts. It is conceivable that MP has the potential to bind actin in protoplasts or in the presence of a fixative. It is also possible that the binding to plant actin is a property of nonfused MP, although nonfused MP did not show interactions with actin filaments when expressed in mammalian cells (Ferralli et al., 2006). Nevertheless, our findings are consistent with the model that MP and MP/vRNA particle trafficking is primarily ER mediated and does not involve direct interactions with actin filaments.
Although MP:RFP binding along ABD2:GFPlabeled actin filaments was not observed, we noticed that MP:RFP-containing bodies occurred in the vicinity of the filaments. This finding presumably reflects the previously reported association of viral replication complexes and fluorescent protein-tagged 126-kD replicase protein with microfilaments labeled with fluorescent protein-tagged talin or ABD2 markers (Liu et al., 2005).
Collectively, our results suggest that the PD-targeted transport of MP and associated viral RNA occurs by diffusion in the ER membrane and that TMV movement can be inhibited by conditions that dominantly block this pathway (Fig. 7). The actomyosin system, including its actin-binding factors, may control this pathway by supporting or slowing down the transport of membrane-embedded protein complexes.

DNA Constructs
The plasmid encoding the cDNA of TMV-MP:RFP has been described previously . Binary plasmids pK7.ABD2:GFP and pK7. ABD2:RFP were generated by Gateway cloning. Here, the open reading frame of ABD2 was PCR amplified from pRSAT.GFP:ABD2 (Sheahan et al., 2004) and cloned into Gateway donor vector pDONR zeo (Invitrogen). Subsequently, the open reading frame was recombined into destination plasmids pK7FWG2.0 (Karimi et al., 2002) and pK7RWG2.0 (kindly provided by R. Tsien) for expression in C-terminal fusion to GFP and RFP, respectively. Binary plasmids encoding GmMan1:tdTomato, GFP:HDEL, and RFP:HDEL were provided by Prof. A. Nebenfü hr. The membrane marker TM17:GFP was provided by N. Paris.
For visualization of actin filaments, the cells were briefly stained for 2 min with 5 mg mL 21 rhodamine-phalloidin in staining buffer (50 mM PIPES, 5 mM EGTA, 2 mM MgCl 2 , 0.05% Triton, and 5% dimethyl sulfoxide) and immediately observed by fluorescence microscopy.

Plant Material
Nicotiana benthamiana wild-type plants as well as plants of transgenic plant line 16c (Ruiz et al., 1998) and of a plant line transgenic for ABD2:GFP (see below) were grown and maintained under greenhouse conditions (50% humidity, 18°C-25°C). Three-to 4-week-old plants were used for infiltration and inoculation experiments.

Agroinfiltration
Transformed agrobacteria (Agrobacterium tumefaciens, line LBA4404; Life Technologies) were grown at 28°C in 2 mL of Luria-Bertani medium containing antibiotics. Upon harvest by centrifugation and resuspension in water at an optical density at 600 nm of 0.05, bacteria were syringe infiltrated into the abaxial side of the leaf.

Plant Transformation
ABD2:GFP transgenic N. benthamiana plants were generated by Agrobacterium-mediated leaf disc transformation (Horsch et al., 1985) using plasmid pK7.ABD2:GFP. Following recovery of whole plants on selective medium, plants were screened for expression of ABD2:GFP using fluorescence microscopy. Four different lines showing similar ABD2:GFP localization patterns to filaments were selected for further experiments.

LatB Treatment
LatB solutions (10 or 100 mM in BY-2 cell medium) were freshly prepared for each experiment and syringe infiltrated into the abaxial leaf side. Control infiltrations were performed with the same buffer but without inhibitor.

Western-Blot Analysis
Total proteins were isolated from leaf discs (diameter, 0.5 cm) in extraction buffer (50 mM Tris-HCl, 100 mM dithiothreitol, 10% glycerol, 2% SDS, and 0.1% bromphenol blue). Following denaturation at 95°C for 5 min and subsequent sonication for 20 min in a Branson 2200 sonicator, the proteins were size fractionated by SDS gel electrophoresis, blotted onto membranes, and probed with a 1:2,000 dilution of mouse monoclonal GFP antibody (Clontech) and anti-mouse IgG horseradish peroxidase-linked secondary antibody (Molecular Probes). The antibody-decorated membranes were then treated with the Lumi-Light PLUS Western Blotting Kit (Roche) for chemiluminescence signal detection autoradiography.

Confocal Laser Scanning Microscopy
Confocal laser scanning microscopy was performed using a Zeiss LSM510 laser scanning confocal microscope with a C-Apo-chromat (633/1.2-W Korr) water objective lens in multitrack mode. Excitation/emission wavelengths were 488 nm/505 to 545 nm for GFP and 543 nm/585 to 615 nm for RFP. Images were acquired using LSM510 version 2.8 software (Zeiss).

Time-Lapse Fluorescence Microscopy
Time-lapse fluorescence microscopy was performed using a Nikon TE2000 inverted microscope equipped for real-time imaging with a Roper CoolSnap digital CCD camera, piezo-driven Z-focus, and a 603, 1.45 numerical aperture total internal reflection fluorescence objective. Excitation/emission wavelengths were 460 to 500 nm/510 to 560 nm for GFP and 550 to 600 nm/615 to 665 nm for RFP. For simultaneous dual-color acquisitions, a Dual-View beam splitter (Optical Insights) was used. The beam splitter was equipped with GFP-mRFP1 exciter and mirror as well as with emission filters HQ510/30m for GFP and HQ650/75m for RFP. To reduce red background emission fluorescence caused by chlorophyll, an e640sp short-pass filter (Chroma) was inserted between the beam splitter and the CCD camera. Images were acquired using Metamorph (6.2r6) software.

Image Processing
Following acquisition, images were processed using Metamorph (6.2r6), ImageJ (1.38u), and Adobe Photoshop (version 7.0) software. Specific algorithms (http://rsb.info.nih.gov/ij/macros/) implemented in the software ImageJ (1.38u) were used for dynamic pixel analysis and display ("slice-toslice-difference"), the tracking of individual Golgi stacks and display of paths ("trace"), and the measurement of the sizes of TMV-MP:RFP infection sites ("threshold" to define regions of interest, "particle analyzer" to count the number of pixels within the regions of interest).

Analysis of TMV-MP:RFP Infection Sites
Leaf discs carrying TMV-MP:RFP infection sites were excised at 3.5 dpi and images were taken. The leaf discs were then incubated individually for 24 h in wells of a 24-well plate containing BY-2 cell medium. After the incubation period, images of the same infection sites, applying strictly constant acquisition conditions, were again taken. The fluorescent area of the individual infection sites before and after the 24-h incubation period was measured using ImageJ software. To compare the sizes of many infection sites with different initial sizes, for each infection site the initial size value was set to 1 and the relative fold size increase was calculated. Measurements were performed with multiple infection sites in three individually performed experiments.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Subcellular localization of MP:RFP within TMV-MP:RFP infection sites at 5 dpi.
Supplemental Movie S1. Dynamic movements of actin filaments in epidermal cells transgenic for ABD2:GFP; time is in seconds.
Supplemental Movie S2. Absence of dynamic actin filament movements in epidermal cells upon transient expression of ABD2:GFP; time is in seconds.
Supplemental Movie S3. Dynamic ER movements in epidermal cells transgenic for ABD2:GFP; time is in seconds.
Supplemental Movie S4. Aberrant ER structure and absence of dynamic ER movements in an epidermal cell transiently expressing ABD2:RFP; time is in seconds.
Supplemental Movie S5. Absence of dynamic ER movements but presence of an otherwise normal ER network structure in an epidermal cell transiently expressing ABD2:RFP; time is in seconds.
Supplemental Movie S6. Dynamic Golgi movements in epidermal cells transgenic for ABD2:GFP; the Golgi complexes are labeled with GmMan1:tdTomato; time is in seconds.
Supplemental Movie S7. Absence of dynamic Golgi movements in epidermal cells upon transient expression of ABD2:GFP; time is in seconds.