Viruses use and subvert host cell mechanisms to support their replication and spread between cells, tissues and organisms. Microtubules and associated motor proteins play important roles in these processes in animal systems, and may also play a role in plants. Although transport processes in plants are mostly actin based, studies, in particular with Tobacco mosaic virus (TMV) and its movement protein (MP), indicate direct or indirect roles of microtubules in the cell-to-cell spread of infection. Detailed observations suggest that microtubules participate in the cortical anchorage of viral replication complexes, in guiding their trafficking along the endoplasmic reticulum (ER)/actin network, and also in developing the complexes into virus factories. Microtubules also play a role in the plant-to-plant transmission of Cauliflower mosaic virus (CaMV) by assisting in the development of specific virus-induced inclusions that facilitate viral uptake by aphids. The involvement of microtubules in the formation of virus factories and of other virus-induced inclusions suggests the existence of aggresomal pathways by which plant cells recruit membranes and proteins into localized macromolecular assemblies. Although studies related to the involvement of microtubules in the interaction of viruses with plants focus on specific virus models, a number of observations with other virus species suggest that microtubules may have a widespread role in viral pathogenesis.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Viruses are obligate parasites that encode a limited number of genes, and require complementing host cell functions for replication and spread. Importantly, once viruses enter into cells they require cytoplasmic transport to reach specific subcellular sites for replication, and then subsequent transport of the progeny virus in order to exit the infected cell. Numerous studies have revealed that retrograde and anterograde transport of animal viruses is powered by microtubules and their associated dynein and kinesin motor proteins (Dohner et al., 2005; Dodding and Way, 2011). Thus, viruses have adapted to existing mechanisms for the microtubule-based transport of macromolecules and organelles that is essential for the animal cell to function. However, although plants have a dynamic interphase microtubule cytoskeleton and a particularly large kinesin motor family, it is generally accepted that the transport of macromolecules and organelles in plant cells is driven by myosins moving along actin filaments (Avisar et al., 2008; Griffing, 2010; Tominaga and Nakano, 2012). Nevertheless, certain instances show that also microtubules can play a role in macromolecular and organelle transport in plants (Cai and Cresti, 2012). In pollen tubes, for example, several kinesin motors were found in association with microtubules and organelles like mitochondria, Golgi bodies and vesicles, thus apparently functioning in the organization of microtubules and in the transport of membrane structures (Cai and Cresti, 2010). Moreover, there is evidence indicating that the two cytoskeletal systems – microtubules and microfilaments – interact and share certain functions. Although plants use the actin rather than the microtubule cytoskeleton to remodel the endoplasmic reticulum (ER), there is extensive ultrastructural and functional evidence for the association of the cortical ER network with microtubules in algal and plant cells (Foissner et al., 2009; Griffing, 2010; Hamada et al., 2012). Moreover, whereas organelle motility is microfilament based, microtubules appear to represent specific sites at which organelles often pause in their movement (Chuong et al., 2005; Crowell et al., 2009; Gutierrez et al., 2009; Hamada et al., 2012), or at which they may be positioned (Cai and Cresti, 2012). As will be described in this review, this division of labor may also be reflected in the interaction of plants with viruses that undergo functional contacts with both cytoskeleton systems during their life cycle.
The minimal set of proteins encoded by plant viruses includes proteins required for replication, specialized movement proteins (MPs) for cell-to-cell transport through the gatable plasmodesmata, and coat proteins for encapsidation and protection of the virus genome. Studies to elucidate the cellular mechanisms facilitating the intracellular and intercellular transport of plant viruses were pioneered with Tobacco mosaic virus (TMV). The MP of this virus was the first identified and was shown to form a ribonucleoprotein complex with single-stranded nucleic acids, and to alter the size exclusion limit (SEL) of plasmodesmata (Deom et al., 1992). Subsequently, it was demonstrated that this protein tightly interacts with microtubules as well as with the nearby ER network (Heinlein et al., 1995, 1998a). The plant ER is associated with actin filaments that propel macromolecular transport along the membrane with the help of myosin motor proteins (Sparkes et al., 2009; Griffing, 2010). Consistent with these findings, the studies further detailed below indicate specific roles of the ER, the ER-associated microfilaments and the nearby microtubules in the establishment of ER-associated replication sites, and in the ER-mediated cell-to-cell movement of TMV. As will be discussed in this review, microtubules seem to play a role in the specific anchorage and positioning of viral replication complexes (VRCs) at the cortical ER and in their subsequent release for microfilament-supported transport along the membrane. Although the exact function of microtubules in this process is not yet clear, there are now several other viruses for which an interaction with microtubules has been documented. Microtubules can also play an important role in the formation of specific virus-induced inclusions, facilitating virus transmission between plants by insect vectors, as has been studied in detail for Cauliflower mosaic virus (CaMV). Collectively, the interactions of plant viruses with microtubules observed so far are consistent with a role of cortical microtubules in the formation, coordinated trafficking and positioning of macromolecular complexes in the cortex of plant cells.
Interaction of TMV with the plant cytoskeleton
The spread of TMV within infected plants depends on the replication of the viral RNA genome (vRNA) in each infected cell and on the successful transport of the viral genome into the adjacent cells via plasmodesmata (Maule, 2008; Peña and Heinlein, 2012). Both processes depend on intricate interactions between the virus and its host. The vRNA encodes four major protein products, the 126- and 183–kDa replicase proteins, the 30–kDa MP and the 17.5–kDa coat protein (CP). The larger 183–kDa replicase protein is produced by read-through of an amber Stop codon that terminates the translation of the 126–kDa protein. In its extended C terminus the protein contains the RdRp domain required for virus replication. The smaller replicase protein acts as a suppressor of RNA silencing through binding and sequestration of small RNAs (Csorba et al., 2007; Kurihara et al., 2007; Vogler et al., 2007; Hu et al., 2011). The two replicase proteins interact with each other and are found in replication complexes isolated from infected plants (Osman and Buck, 1996; Watanabe et al., 1999; Goregaoker et al., 2001), suggesting that vRNA replication and its shielding away from the host RNA silencing machinery are coupled processes. The virus replicates within VRCs associated with the ER (Heinlein et al., 1998a; Asurmendi et al., 2004). The ER membrane network is continuous between cells through plasmodesmata (Ding et al., 1992; Maule, 2008), and provides a direct pathway for the spread of replicated virus from the replication sites in infected cells into the ER network of non-infected cells (Figure 1; Niehl and Heinlein, 2011). The MP of TMV facilitates the cell-to-cell passage of the infectious particle by forming a ribonucleoprotein complex with the vRNA (Citovsky et al., 1990), and by increasing the size exclusion limit of plasmodesmata (Wolf et al., 1989). The CP is dispensable for the cell-to-cell movement of the virus (Holt and Beachy, 1991), indicating that TMV moves its genome in a non-encapsidated form, probably by exploiting mechanisms that support the cell-to-cell and systemic trafficking of endogenous RNA molecules (Hofmann et al., 2007). A number of observations suggest that in addition to the MP, the 126–kDa replicase protein is also involved in cell-to-cell movement (Holt et al., 1990; Derrick et al., 1997; Hirashima and Watanabe, 2001; Knapp et al., 2001, 2005). This finding may indicate a role of silencing suppression in the successful movement of the vRNA, but is also in agreement with observations suggesting that the virus moves between cells in the form of intact VRCs (Kawakami et al., 2004).
Consistent with the tight association of ER membranes with microfilaments there is evidence for a role of the actin cytoskeleton in facilitating the targeting of MP to plasmodesmata (Wright et al., 2007), in the development and subcellular movements of VRCs (Liu et al., 2005), as well as in the spread of infection (Kawakami et al., 2004; Liu et al., 2005; Harries et al., 2009; Hofmann et al., 2009). The inhibition of virus movement in infection sites was observed upon the inhibition of the actin-myosin system after treatment with either latrunculin B or cytochalasin D, or by the silencing of actin or of specific myosins (Kawakami et al., 2004; Liu et al., 2005; Harries et al., 2009). Interestingly, unlike in these experiments in which the actin-myosin system was inhibited for several days during or in advance of the observations, infection sites continued to expand when the actin cytoskeleton was degraded with latrunculin B for only 24 h (Hofmann et al., 2009). Although the inhibitor may not be sufficiently able to eliminate all essential actin targets during this time, this latter finding may suggest that microfilaments and their motors have a supportive rather than a direct role in TMV movement. This proposal is in agreement with the ability of the ER to still support the lateral diffusion of macromolecular complexes upon actin degradation, albeit with reduced efficiency (Sparkes et al., 2009; Griffing, 2010). The inhibition of virus movement by the prolonged inhibition of the actin cytoskeleton is consistent with ER-mediated transport, as the normal structure of the cortical ER network can be strongly affected under these conditions (Wright et al., 2007; Hofmann et al., 2009). One might wonder whether the lack of inhibition of virus movement during the inhibition of the actin cytoskeleton for 24 h (Hofmann et al., 2009) argues against a role of the actin cytoskeleton in facilitating ER-mediated virus movement; however, this is not likely to be the case. Given that the spread of infection depends on the successful movement of only a very few virus particles (Sacristan et al., 2003; Li and Roossinck, 2004; Gonzalez-Jara et al., 2009; Gutierrez et al., 2012), the spread of infection may continue even though the disruption of actin may drastically reduce the number of virus genomes that reach the adjacent cells. Instead, it is likely that the ER-associated actin cytoskeleton and its motors indeed facilitate ER-mediated virus movement, as virus movement and also the intracellular myosin-driven movements of Golgi complexes along the ER were inhibited upon the overexpression of an actin-binding protein (the actin-binding domain 2 of fimbrin fused to GFP, ABD2:GFP). Importantly, virus movement was restored upon disrupting the inhibitory actin complex with latrunculin B (Hofmann et al., 2009). Together, these observations strongly support the view that virus trafficking is facilitated, but does not depend on, the ER-associated actin network and its motors.
A role of actin filaments in the cell-to-cell movement of viruses by controlling the SEL of plasmodesmata is suggested by the ability of the MPs of TMV, and of Cucumber mosaic virus, to sever actin filaments in vitro, and that the MP-induced increase in the plasmodesmal SEL depends on this activity, and can be inhibited by co-treatment of the plants with the actin filament-stabilizing agent phalloidin (Su et al., 2010). However, it remains to be seen whether the implied reorganization of the actin cytoskeleton that leads to an increase in SEL indeed occurs inside plasmodesmata, and is directly responsible for changing the permeability of plasmodesmata. Although there is immunological and other evidence for the association of actin and myosin with plasmodesmata, a functional role of these components in the regulation of the SEL is not easy to demonstrate. Cytoskeleton-modifying treatments are usually applied through whole tissues, and even if microinjected may create effects on cell metabolism and physiology in addition to specific effects on plasmodesmata (White and Barton, 2011). Moreover, although the actin-myosin model for regulating the SEL is attractive, it remains questionable how the molecular size dimensions of actin and myosin would fit into the narrow channel (Burch-Smith and Zambryski, 2012). Unlike initial suggestions from immunostaining experiments in protoplasts (McLean et al., 1995), the ability of MP to support the intercellular spread of the TMV genome in infected leaves does not seem to involve direct interactions of the protein with actin filaments in the cytoplasm (Figure 2h,i; Hofmann et al., 2009). Further studies may demonstrate whether the protein interacts with actin filaments associated with plasmodesmata.
In addition to these observations that imply potential functions of the actin cytoskeleton in virus movement by facilitating virus transport along the ER, and in controlling the SEL of plasmodesmata, there are numerous experimental findings indicating that the microtubule cytoskeleton also plays a role. These findings have shown that MP interacts with microtubules and that this interaction is directly or indirectly coupled with MP function in virus movement. As will be further described below, MP function in virus movement appears to involve the association of MP with distinctly localized, microtubule-associated events by which movement-competent viral complexes are formed and guided along the ER.
The TMV movement protein interacts with microtubules
The interaction of TMV MP with plant microtubules became immediately evident when the MP was expressed in fusion with GFP during infection and its subcellular localization could be observed in vivo (Heinlein et al., 1995). Indeed, in cells behind the front of infection formed by TMV-MP:GFP in leaves (Figure 2a,f,g), and also in infected Nicotiana tabacum (tobacco) BY–2 protoplasts (Figure 2b,d,e), the protein accumulates to high levels and aligns along the filaments (Heinlein et al., 1995, 1998a; Padgett et al., 1996; Boyko et al., 2000b). The ability of MP to bind microtubules is independent of infection, and is also observed upon ectopic MP expression, for example in transfected BY–2 protoplasts (Heinlein et al., 1998a), in transgenic BY–2 suspension culture cells (Boutant et al., 2009) or in transfected mammalian cells (Figure 2c; Boyko et al., 2000a; Ferralli et al., 2006). MP even binds to FtsZ, the prokaryotic progenitor of eukaryotic tubulin, as was shown upon inducible expression in Anabaena (Heinlein et al., 1998b). In vitro experiments have demonstrated that recombinant MP purified from Escherichia coli binds to α,β–tubulin dimers as well as to preassembled microtubules, thus indicating that the MP binds microtubules via direct protein–protein interactions (Ashby et al., 2006; Ferralli et al., 2006). MP-aligned microtubules are dynamically and structurally stabilized, and thus resistant against disruption by cold, calcium or microtubule polymerization inhibitors (Boyko et al., 2000a; Ashby et al., 2006; Ferralli et al., 2006). Consistent with direct binding, biochemical analysis of MP:microtubule complexes isolated from infected plant tissues by two-dimensional polyacrylamide gel electrophoresis did not reveal any characteristic change in the pattern of co-purified proteins, suggesting that the MP does not recruit other specific proteins to the complex (J.A. Ashby, J. Hofsteenge, and M. Heinlein, unpubl. data); however, a tobacco microtubule-associated protein termed MPB2C (for MP-binding protein 2C) was found to interact with MP and to play a role in recruiting or fixing transiently overexpressed MP to the polymer. The ectopic co-expression of NtMPB2C with the MP interfered with the ability of MP to move between cells (Kragler et al., 2003). Virus-induced gene silencing of the NtMPB2C homolog in Nicotiana benthamiana reduced the level of microtubule-associated MP but did not affect the ability of MP or of TMV to spread between cells (Curin et al., 2007). Transgenic Arabidopsis lines overexpressing the Arabidopsis homolog of NtMPB2C showed changes in growth and phenotype as well as resistance to Oilseed rape mosaic virus (ORMV; Ruggenthaler et al., 2009), a TMV relative that causes symptoms upon infection in Arabidopsis (Aguilar et al., 1996; Hu et al., 2011). Collectively, these observations indicate that microtubules are indeed part of a pathway directly or indirectly involved in TMV movement between cells, and that MPB2C acts as a regulator of this pathway. This pathway may be of general importance in macromolecular intercellular trafficking, as transient NtMPB2C overexpression also interfered with the cell-to-cell trafficking of KNOTTED, a non-cell-autonomous protein playing important roles in plant development (Winter et al., 2007). Expression of MPB2C also interfered with the cell-to-cell movement of Potato virus X (PVX, see below; Cho et al., 2012).
The interaction between the MP and microtubules is involved in the formation of virus factories
During infection the MP interacts with other cellular targets in addition to microtubules, notably with plasmodesmata and the ER (Figure 3; Heinlein et al., 1998a). The MP is a hydrophobic protein that in differential fractionation experiments behaves like a membrane-integral, or tightly membrane-associated, protein (Reichel and Beachy, 1998). It contains two predicted transmembrane domains (Reichel and Beachy, 1998; Brill et al., 2000, 2004) that are involved in its association with the ER (Fujiki et al., 2006). The MP associates with the ER immediately upon the onset of its expression during early infection. Infection is initiated when a TMV RNA genome enters a cell and is anchored to the ER in a 5′ methylguanosine [m7G(5′)pppG] cap-dependent manner to initiate the replication and translation of the virus (Christensen et al., 2009), leading to the establishment of ER-associated VRCs. Anchorage of the VRCs to the membrane may be enhanced through interaction of the viral 126– and 183–kDa proteins with the transmembrane proteins TOM1 and TOM3, which contribute to the efficiency of TMV replication (Ishikawa et al., 1993; Yamanaka et al., 2000; Hagiwara et al., 2003). The accumulation of MP produced by the ER-localized VRCs (‘early VRCs’) coincides with the formation of large ER inclusions (‘late VRCs’) that contain viral replicase and vRNA in addition to MP, and probably function as virus factories (Heinlein et al., 1998a; Más and Beachy, 1999). In mature form these inclusions/late VRCs may represent the so-called viroplasms or X–bodies described in the classical literature (Esau and Cronshaw, 1967; Hills et al., 1987). Their formation is associated with rearrangements of the ER membrane, that are likely caused by the accumulated MP as the inclusions diminish and reconstitute a native ER structure when MP becomes degraded by the 26S proteasome (Reichel and Beachy, 1998, 2000). The inclusions are aligned along underlying cortical microtubules (Figure 4a–i; Heinlein et al., 1998a), and functional, microtubule-interacting MP is required for their native distribution in the cortical cytoplasm (Más and Beachy, 2000), suggesting that microtubules act as a scaffold, along which the early VRCs are anchored and develop into mature factories. The role of microtubule-interacting MP in the recruitment and rearrangement of ER membranes is also observed upon ectopic MP expression in mammalian cells. Here, the ER membranes become depleted in outer, microtubule-free areas of the cells, and are strongly enriched in regions exactly overlapping with the MP:microtubule complex (Figure 4j–l). Additional experiments have shown that MP-induced membrane recruitment in mammalian cells indeed requires the microtubule association of the MP, and that the microtubule association of the MP is maintained upon the chemically induced vesiculation of the ER (Ferralli et al., 2006). These observations suggest that the interaction of the MP with microtubules has a significant role in membrane recruitment, and in the formation of ER-derived inclusions and virus factories. The analysis in plants of virus mutants harboring deletions in the MP confirmed that the ability of the MP to interact with microtubules is functionally upstream of the ability of the MP to rearrange the ER and to form ER inclusions (Boyko et al., 2000c). Consistent with the association of ER membranes with actin, the inclusions contain microfilaments (Figure 2i; Hofmann et al., 2009; Liu et al., 2005) that may provide local structural support and also serve as tracks for intracellular motility (Liu et al., 2005).
MP forms mobile microtubule-associated particles in cells at the front of infection
Previous studies indicated that the ability of TMV RNA to associate with the ER is independent of translation and is mediated by the RNA itself. Indeed, microinjected fluorescence-tagged TMV RNA immediately formed granules that associated with the ER in a 5′ cap-dependent manner (Christensen et al., 2009), presumably by docking to the ER-localized translation machinery. It appears likely that these observations reflect the events that occur when TMV-RNA moves from an infected cell into an as yet non-infected cell, and associates with the ER to initiate the translation and the formation of VRCs. The leading front cells of spreading TMV-MP:GFP infection sites in leaves contain very small and highly distinct MP:GFP complexes that may represent such early VRCs. The complexes occur in the vicinity of microtubules, and some of them show directional movements along or in the vicinity of the polymer (Figures 3 and 5a–d; Boyko et al., 2007; Sambade et al., 2008). The complexes are also formed in infected tobacco BY–2 protoplasts (Figure 5e; Heinlein et al., 1998a). Their formation is a property of the MP as similar complexes are also observed when fluorescent protein-tagged MP is expressed in the absence of infection in leaves. Under these conditions, the MP associates with its own mRNA, as was shown by labeling the mRNA with MS2 RNA labeling technology (Sambade et al., 2008). Moreover, movie data demonstrate that most of the complexes remain anchored at cortical sites, and that some of them suddenly detach from these sites and exhibit directional stop-and-go movements along the ER (Figure 5f–l). The ER-associated mobile particles pause in contact with underlying microtubules and continue along individual microtubules (or microtubule bundles; Figure 5m–p), or move from one microtubule to the next (Figure 5q–t; Sambade et al., 2008). These stop-and-go movements are very similar to those observed for MP particles/early VRCs in cells at the leading front of infection (Figure 5a–d; Boyko et al., 2007). Interestingly, upon application of the microtubule polymerization inhibitor amiprophos-methyl (APM) the MP complexes remained stably anchored at microtubule sites. Application of latrunculin B, which inhibits actin polymerization, reduced but did not inhibit the particle movements (Sambade et al., 2008). These observations are consistent with MP particle movement along the ER/actin network; however, whereas the actin cytoskeleton and associated motor proteins may accelerate the efficiency of lateral diffusion along the ER, microtubules appear to control the release of the particles from microtubule-associated anchorage sites, as well as the overall direction for movement along the ER (Sambade et al., 2008; Sambade and Heinlein, 2009).
The ability of APM treatment to interfere with MP particle movements suggests that the release of MP particles/early VRCs from anchorage sites depends on induced microtubule polymerization. This is in agreement with other results indicating that the MP interacts with GFP-fused microtubule-tip protein EB1 (Figure 6g–i; Brandner et al., 2008), and with the microtubule-organizing center component γ–tubulin (Sambade et al., 2008). Moreover, when expressed in mammalian cells, the MP interfered with the recruitment of γ–tubulin by the centrosome (Figure 6a), and with centrosomal microtubule nucleation activity, strongly suggesting that MP interacts with the cellular machinery that polymerizes microtubules (Ferralli et al., 2006). During the early infection stages, the MP often localizes to microtubule branch points or intersections (Figure 6b–f), which may exhibit microtubule polymerization activity (Figure 6d–f). Consistent with a role of microtubule polymerization, tobacco mutants that are affected in microtubule polymerization dynamics are compromised in their ability to support efficient virus movement (Ouko et al., 2010). Microtubule polymerization could support virus movement by several potential mechanisms (Figure 7a; Peña and Heinlein, 2012; Sambade and Heinlein, 2009). Whereas the attachment complex release (ACR) mechanism uses microtubule polymerization to release the MP particles/VRCs from attachment/assembly sites, the tip-attachment complex (TAC) mechanism uses microtubule polymerization, thus growing microtubules, to push the complexes along the ER. The TAC mechanism plays an important role in ER tubule motility (Waterman-Storer and Salmon, 1998), and is able to provide strong pushing forces, as shown, for example, by the demonstrated ability of polymerizing microtubules to move whole nuclei through the cytoplasm (Zhao et al., 2012). As will be described further below, microtubule polymerization could also support virus movement by allowing the formation of movement-competent VRCs.
The ability of MP to interact with microtubules is correlated with the spread of infection
In vivo studies using TMV derivatives encoding functional, dysfunctional and temperature-sensitive mutants of MP fused to GFP demonstrated that the ability of MP to interact with microtubules is correlated with MP function in virus movement, thus indicating a role of microtubules in vRNA/early VRC movement in cells at the leading front of infection. Initial studies correlated the increased efficiency by which TMV spreads in infected plants at elevated temperatures with an increase in the microtubule interactions of MP (Lebeurier and Hirth, 1966; Matthews, 1991; Boyko et al., 2000b). Subsequently, it was shown that amino acids 1–213 of the MP, previously shown to be required for virus movement (Gafny et al., 1992), are also required for the interaction of MP with microtubules (Boyko et al., 2000c). Moreover, a dysfunctional MP carrying a three amino acid deletion mutation in the N terminus (MPNT–1, also named TAD1, dMP or MP∆aa3–5 in the literature), and known to confer MP-derived resistance in transgenic plants (Lapidot et al., 1993; Cooper et al., 1995), interfered with the microtubule association of virus-encoded MP during infection (Kotlizky et al., 2001). The ability of the MP to interact with microtubules also correlated with the functional restoration of a dysfunctional amino acid exchange mutation (Pro81Ser) by second-site mutations (Thr104Ile and Arg167Lys). This finding also indicated that MP function and microtubule association depend on a specific fold that allows distant primary sequences and secondary structure elements to interact (Boyko et al., 2002). Interestingly, Thr104 has been identified as phosphorylated residue, and the replacement with non-phosphorylatable and phosphorylation-mimicking mutations indicated a role of this phosphorylation in controlling MP function in virus movement (Karger et al., 2003). Thus, it may be possible that phosphorylation and dephosphorylation of Thr104 represents a functional switch that determines whether or not the MP acquires a specific fold that allows it to interact with microtubules and to function in virus movement.
The strongest line of evidence for a functional implication of microtubules in TMV movement is based on conditional mutations in the MP that confer a movement-deficient phenotype at an increased temperature of 32°C (Jockusch, 1968; Nishiguchi et al., 1978; Ohno et al., 1983). These mutations interfere with microtubule interaction of the MP at the temperature non-permissive for virus cell-to-cell movement, and allow the MP to interact with microtubules at the permissive temperature (Boyko et al., 2000a, 2007). Interestingly, the non-permissive temperature affected the formation of microtubule-proximal MP particles/early VRCs in cells at the leading front of infection, as well as the accumulation of MP along microtubules during later stages. This indicates that the formation of microtubule-proximal MP particles/early VRCs during the early stages of infection and the accumulation of MP along microtubules at late stages reflect functionally related processes. A role of microtubules is also indicated by a combination of antibody labeling and in situ hybridization applied to protoplasts, which showed that vRNA was localized to microtubules in an MP-dependent manner (Más and Beachy, 1999). Moreover, the localization of vRNA along microtubules depended on the competence of the MP to bind microtubules, as vRNA was mislocalized in cells expressing a mutant MP (TAD5; Kahn et al., 1998) that binds vRNA but fails to associate with microtubules (Más and Beachy, 2000).
The collection of mutations employed in the above functional analysis of the MP revealed the importance of specific domains of the MP involved in microtubule association. It appears interesting that the conditional, temperature-sensitive mutations cluster together in a small domain of the MP that bears structural similarity to the M–loop of α-, β- and γ–tubulin (Boyko et al., 2000a). The tubulin M–loop interacts with the N–loop of tubulin molecules in the adjacent microtubule protofilaments, and thus plays an important role in the assembly and stability of microtubules (Nogales et al., 1999). The similarity with the tubulin M–loop may allow the MP to directly bind to free or assembled tubulin of either isoform, including γ–tubulin, and may also identify the MP as a binding target for tubulin co-factors. Indeed, as already mentioned, MP was shown to bind the microtubule-associated protein MPB2C (Kragler et al., 2003) and GFP-fused Arabidopsis end-binding protein 1 (AtEB1a; Brandner et al., 2008). However, whether the M–loop similarity domain is indeed required to bind these factors still needs to be determined.
It should be noted that the targeting of the temperature-sensitive MP to plasmodesmata was not inhibited at the non-permissive temperature (Boyko et al., 2000a). Thus, the targeting of the MP to plasmodesmata is not sufficient for virus movement, and is likely to be independent of the ability of the MP to interact with microtubules. Experiments using fluorescence recovery after photobleaching (FRAP) combined with specific inhibitors demonstrated that the targeting of GFP-tagged MP to plasmodesmata depends on an intact ER network, and is enhanced by an intact actin cytoskeleton, whereas the microtubule cytoskeleton has no effect (Wright et al., 2007). Thus, MP produced in anchored VRCs may target the plasmodesmata via the ER/actin network independently from microtubules, and this may occur before the VRCs themselves are released for movement. Collectively, the above evidence indicates that the ability of the MPs to interact with microtubules and MP function in virus movement are correlated with the formation of MP particles/early VRCs and their controlled release for transport along the ER/actin network in cells at the leading front of infection. This suggests a critical role of microtubule-mediated attachment and release in the formation of movement-competent VRCs.
Accumulation of MP along microtubules during late infection
The role of microtubules changes during the course of infection. During early infection microtubules seem to act as scaffolding for the localized attachment and subsequent release and movement of VRCs along the ER. At later stages of infection, in cells behind the infection front, the microtubules act as a scaffold, along which anchored VRCs grow and mature into inclusions or X–bodies that harbor virus factories and produce virus progeny. At even later stages the microtubules may accumulate MP along their length (Figures 2 and 3). The accumulation of MP along the length of microtubules is not required for virus movement, as this phenomenon usually occurs in cells far behind the leading front of infection (Padgett et al., 1996; Heinlein et al., 1998a). Consistently, downregulating microtubule alignment of accumulated MP by knocking down the expression of MPB2C had no consequence on MP function in virus movement (Curin et al., 2007), and a virus variant (R3) that shows reduced accumulation along microtubules exhibited faster, rather than slower, cell-to-cell movement (Gillespie et al., 2002). The observation of microtubule-aligned MP is indeed variable, and depends on the nature of the host for infection (Padgett et al., 1996), the virus variant under study (Padgett et al., 1996; Heinlein et al., 1998a; Gillespie et al., 2002) and on environmental conditions (Boyko et al., 2000b). Nevertheless, the accumulation of MP along microtubules causes microtubule stabilization (Boyko et al., 2000a; Ashby et al., 2006; Ferralli et al., 2006), and interferes with kinesin motility (Ashby et al., 2006), intracellular MP particle/VRC movements (Boyko et al., 2007) as well as with intercellular virus spread (Curin et al., 2007), and thus may play an important role in prohibiting further virus movement between cells that are already behind the advancing front of virus infection. As the alignment of accumulated MP along microtubules precedes the sudden disappearance of the MP during even later stages, i.e. in the center of infection sites, MP-aligned microtubules may also be associated with a pathway leading to the degradation of the MP by the 26S proteasome (Padgett et al., 1996; Heinlein et al., 1998a; Reichel and Beachy, 2000). This hypothesis is supported by the increased stability of the R3 MP, showing decreased accumulation along the polymer (Gillespie et al., 2002); however, several observations argue against a direct role of microtubules in MP degradation. First, the MP is still degraded in the center of infection sites upon treatment of the leaves with the microtubule-disrupting agent APM (Ashby et al., 2006); second, an MP mutant with increased microtubule binding was more stable than the wild-type MP in infected tobacco BY–2 protoplasts (Kotlizky et al., 2001); and, third, unlike the MP in crude extracts, microtubule-associated MP is not ubiquitinated (Ashby et al., 2006). Our recent studies indicate that the degradation of the MP depends on the AAA ATPase activity of CDC48 (Niehl et al., 2012). This protein is induced by infection, binds and extracts the MP from ER inclusions and allows the degradation of the protein in the cytosol by an ER-associated degradation (ERAD)-like mechanism. Overexpression of the protein enhanced the accumulation of MP along microtubules, which may suggest a role of microtubules in stockpiling the MP before degradation. Recent studies demonstrate that ERAD substrates are de-ubiquitinated for dislocation from the ER (Ernst et al., 2011; Tsai and Weissman, 2011), which could explain the lack of ubiquitination of microtubule-aligned MP (Ashby et al., 2006). Whether microtubule-aligned MP is re-ubiquitinated for subsequent degradation, or whether this fraction of the MP enters yet another pathway before degradation, remains to be investigated.
TMV movement in the presence of microtubule-polymerization inhibitors
The correlation between MP interactions with microtubules and MP function in TMV movement provides substantial evidence for a role of microtubules in the spread of infection; however, when the role of microtubules in virus movement is tested by the treatment of plants with microtubule polymerization inhibitors, contradictory results are obtained. Although the presence of the microtubule polymerization inhibitor APM compromises the movements of individual MP particles/VRCs at the subcellular level within minutes under the microscope (Sambade et al., 2008), infection sites in leaves infiltrated with oryzalin, APM or colchicine continue to expand (Gillespie et al., 2002; Kawakami et al., 2004; Ashby et al., 2006). As already discussed with respect to the lack of inhibition of virus movement by treatment with latrunculin B for 24 h, the most likely explanation for the latter finding is that inhibitor treatments are not sufficient to eliminate all target structures. That the virus still finds intact microtubules in APM-treated leaves has been demonstrated (Seemanpillai et al., 2006), and is consistent with the stabilization of microtubules by microtubule-associated proteins (MAPs), and the MP of TMV is a particularly strongly stabilizing MAP (Ashby et al., 2006; Ferralli et al., 2006). Moreover, even if the inhibitor caused a strong reduction in the number of virus particles that enter the non-infected cell, the spread of infection would be expected to continue normally. As plant cells accumulate enormous quantities of virus progeny (Nixon, 1956), infection may overcome any partial inhibition of essential processes by inhibitor treatments.
It should be noted that although microtubule inhibitors do not interfere with the spread of a replicating virus, the inhibitors are nevertheless useful for testing the role of microtubules in specific subcellular events such as the release of individual MP particles/VRCs from anchorage sites, as noted above. The treatment with the inhibitors is also useful for testing the intercellular transport of non-replicating molecules such as proteins. For example, studies employing the microtubule polymerization inhibitor oryzalin, as well as other conditions that affect the organization or dynamics of microtubules, have recently shown that microtubules are required for the intercellular movement of the SHORT-ROOT transcription factor, and thus for determining endodermal cell fate in the Arabidopsis root (Wu and Gallagher, 2013).
Microtubule-associated MP functions may be linked to an aggresomal pathway
Membrane-associated virus inclusions or factories similar to those formed during TMV infection are common to plant viruses (Laliberte and Sanfacon, 2010; Verchot, 2011; Grangeon et al., 2012), and are also known for virus infection in animal cells (Novoa et al., 2005; Netherton et al., 2007; de Castro et al., 2013). Viral factories in animal cells often occur at pericentriolar sites close to the microtubule-organizing center, where they can benefit from minus end-directed microtubular transport to deliver cellular and viral proteins to facilitate replication and assembly. In doing so, viruses may exploit mechanisms involved in the formation of aggresomes, which involves the microtubule-mediated delivery and immobilization of aggregated proteins for subsequent degradation by proteasomes or autophagy. Thus, replication factories such as those formed by TMV may indeed originate as a consequence of an innate immune or stress response that recognizes the viral components as foreign, overexpressed or unfolded, and uses an aggresomal pathway for their storage and degradation (Wileman, 2006, 2007).
The association of the TMV factories with microtubules, and the formation of the factories from early VRCs, may be consistent with the involvement of such a pathway for the delivery of virus components and the formation of viral factories. In such a scenario, the interaction of MPs with microtubules and with microtubule nucleation and polymerization factors would allow the virus to generate the microtubule scaffolding for subsequent kinesin motor-mediated delivery of the membrane and protein components involved in the aggresomal build-up of the VRC/virus factory.
This idea may be supported by the previously mentioned interaction of MP with CDC48 (Niehl et al., 2012, 2013), which is essential for aggresome formation in yeast (Wang et al., 2009). CDC48 is known to interact with ubiquitination and de-ubiquitination enzymes, and to control the association of ER membranes with proteins (Meusser et al., 2005; Crosas et al., 2006; Rumpf and Jentsch, 2006). This protein has also been proposed to mediate the delivery of proteins to the proteasome or the aggresome, respectively, depending on the proteostatic state of the cell (Ju and Weihl, 2010). Based on the finding that MP interacts with CDC48, it seems possible that the whole course of TMV infection in a given cell may rely on an aggresomal pathway. This pathway could provide the machinery for the initial recruitment and concentration of virus- and host-encoded factors required for replication, movement and protection against host defenses, and also for the subsequent disposal of these factors after the viral life cycle is complete.
According to the model shown in Figure 7b, VRCs attach at microtubule-associated attachment sites on the ER membrane. During early infection, when only limited quantities of the MP are produced, the proteasome is not saturated and the MP produced in the microtubule- and ER-asssociated VRCs can perform its function in provoking microtubule polymerization and in liberating VRC transport before it is extracted with the help of CDC48 and finally degraded. As the newly polymerized microtubules are attached with their minus ends to the VRC, they may allow the delivery of host factors required for the maturation of a movement-competent VRC or subcomplex with the help of microtubule minus end-directed kinesins. The MP degradation process facilitated by CDC48 may support the formation of the movement-competent VRC by extracting microtubule-associated MP from the VRC, and thus by releasing the complex from its microtubule anchorage. Later, when MP accumulates to high levels and the proteasome pathway becomes saturated, the aggresome, consisting of aggregated MP, viral RNA, ER membrane and presumably other viral and host factors, is formed, and the further recruitment of membranes and proteins along microtubules allows the aggresome to grow and to turn into a viral factory. As CDC48 continues to extract MP while the proteasome pathway is still saturated, the MP may slowly accumulate on microtubules and prevent further recruitment of membranes and proteins, thereby limiting further growth of the aggresome. At final stages of infection, MP expression ceases and the MP and other factors are finally degraded, leading to the shrinkage and final disappearance of the aggresome (in a process that may involve autophagy) and the return of the ER to its pre-infection morphology.
Thus the MP, as an aggregation-prone protein with the capacity to associate with both ER membranes and microtubules, might subvert an aggresomal pathway and associated cytoskeleton and degradation pathways, to regulate the formation and subsequent release or growth of VRCs, followed by their subsequent destruction. In interacting with such an aggresomal pathway the protein would exert a fundamental role in controlling the formation of viral replication factories, as well as of viral intra- and intercellular spread during the course of infection.
Interaction of Other Viruses With Microtubules
Although studies with TMV suggest a role of microtubules in relation to MP functions during infection, studies investigating other plant viruses have excluded a role of microtubules during infection. Usually, this conclusion is based either on the absence of a visible association of the respective viral MP with microtubules or on the observation of continued cell-to-cell spread of infection in the presence of microtubule polymerization inhibitors. However, as we have learned from TMV, both of these observations may not allow final conclusions to be drawn. Many studies also restrict their analysis to the plasmodesmal targeting of ectopically expressed MPs. Here, the absence of an effect of microtubule inhibitors is sometimes stated as a difference to TMV; however, this argument may be incorrect since the MP of TMV does not need to bind to microtubules to accumulate at plasmodesmata (Boyko et al., 2000a, 2007), and, consistent with this, microtubule inhibitors have been shown to have no clear effect in this process (Wright et al., 2007). Of course, virus evolution has evolved multiple different approaches for transmitting infections from cell to cell (Niehl and Heinlein, 2011), and certainly there is the possibility that these do not always involve the microtubule cytoskeleton. Nevertheless, the involvement of microtubules during virus infection may be more widespread than is currently thought. At present, there are already some viruses in addition to TMV for which interactions with microtubules have been described and which deserve further analysis. The MP of Tomato mosaic virus Ob shows microtubule associations that are very similar to those of the MP of TMV, although the MP of this tobamovirus shares only 54% amino acid sequence similarity with the MP of TMV (Padgett et al., 1996). Whether other tobamovirus family members also interact with microtubules remains to be tested. Outside the tobamovirus family, the currently best example for microtubule interactions is given by Potato mop-top pomovirus (PMTV). This virus encodes three ‘triple-gene-block’ (TGB) MPs that act in a coordinated manner to facilitate cell-to-cell and long-distance virus movement, whereby TGB2 and TGB3 act to assist the targeting of TGB1 to plasmodesmata (Erhardt et al., 1999, 2000; Lawrence and Jackson, 2001; Zamyatnin et al., 2004; Lim et al., 2008, 2009). Infection of N. benthamiana plants with a PMTV reporter construct expressing TGB1 fused to YFP revealed that this protein localizes to plasmodesmata in cells at the leading front of infection. In cells just behind the leading front the protein localized to motile bodies moving in association with the ER network and, in the second to fourth cell layer behind the leading front of infection, the protein accumulated along microtubules (Wright et al., 2010). These localization patterns of TGB1 bear striking similarity to those seen previously for the MP of TMV. The role of microtubules in PMTV movement is supported by the observation that an N–terminal deletion in TGB1 that affected its function in virus movement also affected its accumulation along microtubules (Wright et al., 2010); however, because PMTV cell-to-cell movement was not inhibited in the presence of colchicine, the authors concluded that microtubules are dispensable. By using methods other than drug treatment, e.g. by employing plants with modulated microtubule polymerization rates, a functional role of the association of TGB1 with microtubules may yet be revealed by further studies.
Microtubules may also play a role during infection with PVX. Like PMTV, PVX encodes TGB MPs, but in addition this virus requires CP for movement (Chapman et al., 1992; Morozov and Solovyev, 2003). The CP and whole virion particles of this virus were reported to bind to taxol-stabilized microtubules, and to compete with MAP2 binding in vitro (Serazev et al., 2003). A more recent report indicates that the movement of PVX in N. benthamiana is inhibited upon overexpression and enhanced by silencing of NbMBP2C (Cho et al., 2012), which suggests that microtubules have a role in PVX movement. Another viral protein reported to bind microtubules in vitro is the 65–kDa Hsp70 homolog (hsp70 h) of Beet yellows virus, one of the five viral proteins required for movement of this virus (Karasev et al., 1992); however, when transiently expressed in fusion with GFP or mRFP in agroinfiltrated cells, the 65–kDa protein targets plasmodesmata in a microtubule-independent and actin-dependent manner (Prokhnevsky et al., 2005).
Another virus potentially interacting with microtubules is Grapevine fanleaf virus (GFLV). Unlike TMV, this virus moves in the form of virions through tubules assembled by its MP within plasmodesmata. The docking of this MP at plasmodesmata depends on the interaction with plasmodesma-localized PDLP1 that itself requires the secretory pathway for plasmodesmata targeting (Amari et al., 2010). Interestingly, the correct targeting of the GFLV MP to plasmodesmata in BY–2 suspension cells was abolished in the presence of the microtubule polymerization inhibitor oryzalin, whereas the presence of the actin inhibitor latrunculin B had no effect (Laporte et al., 2003). It will be interesting to see by which mechanism the microtubules allow this MP to identify its correct targeting address.
Finally, a yeast two-hybrid screen identified ER- and microtubule-associated proteins as interactors of proteins encoded by Sonchus yellow net rhabdovirus. The results suggest a role of an ER- and microtubule-associated complex in the cell-to-cell transport of this virus (Min et al., 2010).
Role of Microtubules in Virus Transmission by Insects
In addition to viral factories, plant viruses may cause the production of other types of inclusions. Microtubules play a role in the formation of specific electron-lucent inclusion bodies (ElIBs) that are formed in addition to electron-dense inclusion bodies (EdIBs) during infection with CaMV, and which function in the plant-to-plant transmission of the virus by aphids (Figure 8; Bak et al., 2012; Khelifa et al., 2007; Martiniere et al., 2009a). The uptake of the virus by aphids depends on the formation of a transmissible complex that is composed of the virus particle, virus particle-associated P3 and the helper protein/aphid transmission factor P2. P2 mediates the binding of the virus particle to the vector by serving as a molecular bridge. The formation of ElIBs and the recruitment of the viral proteins P2 and P3 into ElIBs rely on an intact microtubule cytoskeleton, as was shown by specific drug treatments. P2 is a microtubule-binding protein (Blanc et al., 1996) able to bind to the polymer in the absence of other proteins, and mediating microtubule association and accumulation of P3 in EllBs (Martiniere et al., 2009a). During the course of infection, both proteins are first seen in the EdIBs, in which the translation of all viral proteins is localized. Before they accumulate in ElIBs, the proteins are seen in association with microtubules, indicating that microtubules are involved in the translocation of P2 and P3 from EdIBs to ElIBs. As the formation of ElIBs was not inhibited by taxol, the microtubules may transport P2 and P3 with the help of microtubule motor proteins. Consistently, an atypical kinesin TBK5 and its Arabidopsis orthologue were found to co-localize with P2 on microtubules as well as in ElIBs, and thus may play a role (Martiniere et al., 2009b). The ElIB forms as a single body in the infected cell that may be related to an aggresome. To facilitate virus uptake by aphids, the ElIB has been proposed to burst open and thus to distribute the transmissible complex in the cell (Bak et al., 2012). Indeed, recent studies have shown that the ElIB or ‘transmission body’ (TB) reacts instantly to the presence of the aphid vector by a rapid and reversible redistribution of its components, including virus particles, onto microtubules and throughout the cell (Martinière et al., 2013). This finding corroborates the importance of the microtubule cytoskeleton in CaMV transmission by aphids, and also illustrates the ability of viruses and infected cells to rapidly respond to cues from the environment.
Microtubules may also play a role in the transmission from insect vector to plants. For example, Rice gall dwarf virus particles replicate and accumulate adjacent to microtubules in cultured insect vector cells. Disruption of microtubules with drugs suppressed the association of viral particles with microtubules, and prevented the release of viruses from cells into the medium, without affecting viral multiplication (Wei et al., 2009).
Although it seems obvious that TMV and several other viruses interact with microtubules, and that microtubules are involved in the cell-to-cell progression of TMV infection, further work is required to exactly define the specific roles of the filaments and associated motors in this process. Collectively, the evidence suggests that the microtubule system acts in concert with the ER network and associated microfilaments. Whereas the ER/actin network provides the fluid medium for VRC transport through the cytoplasm and plasmodesmata, microtubule-associated processes appear to coordinate this process by controlling the assembly and release of movement-competent VRCs, their guidance along the ER membrane network and the development of anchored VRCs into viral production factories. The interactions of TMV with microtubules are mediated by its MP, which binds to microtubules and also to proteins that regulate microtubule nucleation and assembly. Experiments in MP-transfected mammalian cells demonstrated that the MP interferes with the centrosomal recruitment of γ–tubulin, suggesting that MP may sequester γ–tubulin ring complexes (γ–TuRCs) (Figure 6a). Thus, an early VRC that initially binds to the ER membrane and starts to produce MP may act as a microtubule nucleation center with the help of MP (Figure 7b). The establishment of new microtubules with their minus end attached to the VRC may create an aggresome-like situation, in which minus-end directed microtubule motors deliver cargo towards the microtubule nucleating center. The MP present in the VRC thus allows the VRC to mature into a movement-competent complex, or to become a virus factory. The formation of movement-competent VRCs and of the virus factories in association with the ER/actin network and microtubules is likely to depend on the ability of the MP to act as a microtubule-to-ER membrane linker. Interestingly, the critical domain for the microtubule association of MP overlaps with the transmembrane domain of this protein, suggesting that the ability of this protein to interact with membranes or microtubules depends on specific folds. As the MP acts as an oligomer (Brill et al., 2004; Boutant et al., 2010), the protein may form a higher-order structure that combines the different folds into one complex (Figure 7d), and thus allows MP to recruit and stabilize ER membranes along microtubules and towards the VRC. However, whether the MP is indeed critical for viral factory formation requires further study. As the MP is dispensable for virus replication (Meshi et al., 1987), and complexes competent for efficient TMV replication thus form independently of MP, it will be interesting to determine the nature of VRCs and viral factories formed with and without MP. Further studies should also determine whether the formation of virus factories and virus movement depend on kinesins. Unfortunately, current research on the role of microtubules in plant virus infection is still centered on TMV. Future research may also reveal a role of microtubules in the context of other plant viruses, and the recent finding of microtubule association of the TGBp1 of PMTV may be a first step in this direction.
Future investigations may also provide further important insights into the role of CDC48. This protein is a central component of aggresomes, and may have central roles in the regulation of the TMV infection cycle by extracting MP from the ER for degradation (Niehl et al., 2013). This process may liberate VRCs from early aggresomes during the early stages of infection (Figure 7b), and may control aggresome/virus factory size during later stages (Figure 7c). MP extracted from the aggresome/virus factory may be degraded directly or may align to the microtubules before degradation, and thus interfere with kinesin movement and further recruitment of membranes to the aggregate (Figure 7c). Indeed, it was shown previously that microtubule-associated MP interferes with kinesin movement along microtubules in vitro (Ashby et al., 2006), and with MP particle/VRC movements in vivo (Boyko et al., 2007).
Given the potential ability of the MP to instruct an aggresomal pathway for the transient formation of multiple microtubule- and membrane-associated viral factories/viroplasms in the cortex of infected cells, it will be interesting to see whether the dynamic cortical microtubule cytoskeleton could have a general role in supporting dynamic, aggresomal mechanisms for the transient establishment of localized macromolecular complexes involved in cell signaling and growth.
The authors acknowledge support through research funding from the Swiss National Science Foundation (31003A_140694) and the Agence Nationale de la Recherche Scientifique (ANR-08-BLAN-0244), as well as through a postdoctoral fellowship grant from the Zurich-Basel Plant Science Center. The authors also thank M. Drucker for discussion and a critical reading of the virus transmission paragraph.