Getting on the right track: Interactions between viruses and the cytoskeletal motor proteins

The cytoskeleton is an essential component of the cell and it is involved in multiple physiological functions, including intracellular organization and transport. It is composed of three main families of proteinaceous filaments; microtubules, actin filaments and intermediate filaments and their accessory proteins. Motor proteins, which comprise the dynein, kinesin and myosin superfamilies, are a remarkable group of accessory proteins that mainly mediate the intracellular transport of cargoes along with the cytoskeleton. Like other cellular structures and pathways, viruses can exploit the cytoskeleton to promote different steps of their life cycle through associations with motor proteins. The complexity of the cytoskeleton and the differences among viruses, however, has led to a wide diversity of interactions, which in most cases remain poorly understood. Unveiling the details of these interactions is necessary not only for a better comprehension of specific infections, but may also reveal new potential drug targets to fight dreadful diseases such as rabies disease and acquired immunodeficiency syndrome (AIDS). In this review, we describe a few examples of the mechanisms that some human viruses, that is, rabies virus, adenovirus, herpes simplex virus, human immunodeficiency virus, influenza A virus and papillomavirus, have developed to hijack dyneins, kinesins and myosins.

way, they convert chemical energy into the mechanical force that is needed to move along with the filament tracks and transport cell components. 6 The movement and function of motor proteins can be regulated by binding to cargoes, 7 posttranslational modifications such as phosphorylation 8 or other accessory proteins. 9

| MTs, dyneins and kinesins
MTs are present in the cytoplasm ( Figure 1A), in which they determine the subcellular distribution of organelles and guide intracellular transport in between them. As MTs extend to the cell periphery, they can also be structural elements of the cilia and flagella that emerge from the cell surface. MTs can rearrange themselves into the mitotic spindle body, which is essential for the correct segregation of chromosomes during cell division. 10 Structurally, MTs are a polymer of tubulin, which is a heterodimer composed by αand β-tubulin, two globular proteins that bind GTP.
Tubulin units associate longitudinally forming protofilaments and the parallel association of 8-19 protofilaments generates the rigid cylindric MTs that have a diameter of 25 nm ( Figure 1B). [11][12][13] The polymerization of MTs initiates at the microtubule-organizing centers (MTOC), which correspond to the centrosomes in animal cells. 14 There, a complex, which is formed by 12-14 γ-tubulin subunits acting as a cap, nucleates the MTs from the minus end and serves as a template for the polymerization. 14 As tubulins are asymmetrical and associate unidirectionally, MTs possess a polarity, with α-tubulin being at the minus end and β-tubulin at the plus end ( Figure 1A). The plus end polymerizes when a new tubulin heterodimer is incorporated and the GTP molecule in the β-tubulin subunit stabilizes the growing MTs. A GTP-tubulin cap stabilizes and allows the plus end to keep growing, while its loss induces disassembly of MTs. The dynamics of MTs, but also their length and stability, is fine-regulated by accessory proteins such as MT-associated proteins and motor proteins. 15 Two superfamilies of motor proteins, dyneins and kinesins, bind to MTs. 10 Dyneins are divided into two classes, cytoplasmic and axonemal. Both classes are structurally similar and they are composed of one to three heavy chains, which possess the motor domain, variable  Figure 1C). 16 Axonemal dyneins include monomers, heterodimers and heterotrimers, with one, two and three motor heavy chains, respectively and are responsible for the movement of cilia and flagella and intraflagellar transport. Cytoplasmic dyneins, that is, dynein 1 and dynein 2, are homodimers that direct the movement of cell components toward the minus end of MTs. 17 Additionally to its role in the dynein motors, dynein light chain DNCL1/DLC1/LC8 is autonomously involved in MT assembly and stability. 18 In humans, there are 45 kinesins, which form homo-or heterodimers or oligomers, and have a conserved motor domain but variable N-and C-terminal domains ( Figure 1D). 19 Apart from the cargo that they transport, kinesins regulate MTs dynamics by inducing destabilizing conformational changes at their ends. 20 Except for kinesin 13 that depolymerizes MT ends, kinesins traffic toward the plus end of MTs through their motor activity (i.e., anterograde movement) or in the case of C1 kinesin family, toward the minus end (i.e., retrograde movement). 20-22

| Actin filaments and myosins
Actin filaments concentrate under the plasma membrane ( Figure 1A).
Via pushing and pulling forces, actin filaments are involved in determining the cell shape and movement, cell surface projections, intracellular trafficking, cell-cell junctions, contractions and cell division. 23 Actin subunits, also called globular actin (G-actin), are polypeptides of 375 amino acids that bind either an ATP or an ADP molecule.
G-actin polymerizes forming a 6 nm helical assembly known as filamentous actin (F-actin; Figure 2A). 24 The nucleation takes place at different locations throughout the cell and is an unstable process that starts with the formation of an actin dimer followed by the addition of another monomer to form a trimer, which can then further polymerize through the addition of new G-actin units. 25,26 As the subunits are asymmetrical and assemble unidirectionally, they provide the filaments with a polarity, with a minus end, which grows slowly, and a plus end, with a faster polymerization rate. 27 Actin-binding proteins, which also include motor proteins, finely regulate actin nucleation, polymerization, depolymerization and severing. 28 The motor proteins that associate with actin filaments belong to the myosin superfamily, which includes 13 classes in humans. All members share a similar overall structure, which comprises a heavy  Figure 2B). 29 Myosins traffic toward the plus end of the actin filaments, with the exception of myosin VI, which appears to be the only full-length myosin moving toward the minus end. 30 Tail-less myosin IXb (amino acid residues 1-1296) also displays minus-end-directed movement, 31 but not the full-length protein, which has a plus-end directionality. 32

| Intermediate filaments
Intermediate filaments are found in both the cytoplasm, in which they connect the inner side of the plasma membrane with the outer surface of the nuclear envelope, and adjacently to the inner face of the nuclear envelope, where they form the nuclear lamina ( Figure 1A The polymerization is distributed around the inner part of the plasma and nuclear membranes and it starts when two monomers associate parallel into dimers, which then interact with other dimers in an antiparallel way, forming symmetrical tetramers. Eight tetramers associate laterally, resulting in a 10 nm protofilament without polarity ( Figure 2C).

| CYTOSKELETON AND VIRUSES
As numerous other cell pathways and processes, the cytoskeleton is also subverted and exploited by intracellular pathogens, including viruses, to promote their propagation. 38 The life cycle of a virus starts upon its entry into the host cell. The delivered viral genetic material is used as a template for the virus replication and transcription, which can take place in the nucleus and/or cytoplasm and ultimately leads to the production of viral proteins. The newly synthesized genomes and structural proteins assemble to form the virions, which are then released extracellularly. It has been shown that viruses interact with the cytoskeleton elements, mostly through motor proteins, to subvert them and promote their internalization, intracellular transport, replication, assembly and/or egression. 39 The complexity of the cytoskeleton and the enormous differences that there are between life cycles of viruses belonging to different virus families, has led to very diverse types of interactions, which, in most of the cases, remain ill-described. In this review, we will illustrate

| Rabies virus
RABV is a neurotropic pathogen that invades the central nervous system (CNS) of mammals and spreads throughout it. RABV causes the fatal rabies disease, which is characterized by an acute mental dysfunction, followed by a coma and ultimately death. 47 RABV is a singlestranded negative-sense enveloped RNA virus with a small linear genome of 11-15 kb, which encodes for five structural proteins: glycoprotein (G), matrix (M), nucleoprotein (N), phosphoprotein (P) and RNA-dependent RNA polymerase (L; Figure 3A). The RABV life cycle starts when G, which forms trimers and is the only protein exposed The first evidence of a direct interaction between RABV and the MTs was obtained by two studies aiming at deciphering the role of P in the RABV life cycle. Using the yeast two-hybrid system, it was uncovered that P specifically binds to the dynein light chain DNCL1 (Table 1). 50,51 This initial finding was confirmed in vivo by coimmunoprecipitation experiments 50 and confocal microscopy. 51 Moreover, virus purification and subsequent analysis by western blot showed that DNCL1 is incorporated into RABV virions. 50 Through serial N-and C-terminal deletions in P and pull-downs, the DNCL1 binding domain (DNCL1-BD) in P was localized to amino acids 139-151. 50,78 Subsequent generations of binding mutants has allowed to reveal that DNCL1 interaction with P protein mediates its incorporation into RABV viral particles. 79 Because DNCL1 is a component of dyneins that are involved in MT assembly and stability, 18 it was initially hypothesized that the P-DNCL1 interaction was mediating RABV binding to MTs to enhance their viral axonal transport, 79 something suggested by a prior investigation in which MT elongation was inhibited with colchicine. 80 Results from studies in mice, however, pointed out that the binding between P and DNCL1 is not essential Overview of the motor proteins (and their classification) described in this review that are hijacked by specific viruses. for axonal transport of RABV, as RABV strains carrying a mutation in DNCL1-BD spread slower and infect fewer neurons, but the resulting neuronal pathology is similar to the one observed with the wild-type virus. 81,82 Because P is a cofactor of the L polymerase, 83 it was then postulated that the P-DNCL1 interaction plays a role in viral transcription.
Poisson and co-workers analysed in vitro the relevance of the DNCL1-BD in the transcriptional activity of P using a luciferase-based assay. They observed similar transcriptional activities of wild-type and mutant P, which indicated that the binding of DNCL1 to P is not essential for viral transcription. 78 In contrast, Tan and colleagues used realtime PCR and by performing experiments in vivo with mice, observed that although not affecting the cell entry, a RABV strain carrying a mutation in the DNCL1-BD displayed an impairment in the viral transcription and replication, in comparison to the wild-type. 84 These results support the previous data arguing that the P-DNCL1 interaction is probably not key in the dynein-mediated axon transport of RABV viral particles, 81,82 and DNCL1 may rather act as a co-factor to ensure the efficient P polymerase activity. Interestingly, a subsequent study shed light on the mechanism through which DNCL1 influences P-mediated transcription.
Using nuclear magnetic resonance, small-angle X-ray scattering and molecular modelling, it was revealed that DNCL1 binds to an intrinsically disordered region in P, restricting the orientation and mobility of both the intrinsically disordered region and its C-terminal domain. 85 This change stabilizes P in a more active conformation that enhances its polymerase activity. 85 This finding explained why the binding between P and DNCL1 is necessary for an efficient transcription, although this latter can still take place but less efficiently in the absence of the interaction with DNCL1, as both studies concluded. 78,84 In parallel, Bauer and colleagues also described a direct interaction between DNCL1 and the L polymerase (Table 1) and by generating mutant virus strains they identified the DNCL1-BD in L at the amino acid positions 1079-1083. 52 Using a RABV strain with a mutation in the DNCL1-BD of L and confocal microscopy, it was shown that the L-DNCL1 interaction is involved in the association of L with MTs, but also the posttranslational modification and reorganization of MTs. Moreover, the DNCL1-BD of L, together with the DNCL1-BD of P, regulates DNCL1 cellular levels and plays a role in RABV transcription. 52 By measuring the effect of the DNCL1-BD on L using a luciferase-based activity transcription assay, however, another study indicated that DNCL1 binding to L does not regulate its polymerase activity. 85 Thus, future studies are needed to address the precise role of DNCL1-L interaction in the RABV life cycle.
Because G is on the surface of RABV viral particles, 86 it has also been explored whether G is involved in the virion transport within the cells of the CNS using in vitro and in vivo murine models. In particular, engineered recombinant viruses that contain modifications in the structure of the virus envelop have allowed to show that G is required for the optimal axonal MT-mediated retrograde transport of RABV particles. [87][88][89] These results suggest the existence of an interaction between G and dynein, which drives the axonal transport of virions.
However, there is no evidence yet of a direct interaction between G and MTs components.
In conclusion, while DNCL1 interacts with both P and L, only its binding to P regulates the transcription activity of the PL complex.
Why binding to MTs is required to enhance viral transcription is unknown, as well the significance of the DNCL1-L interaction. It also remains to be understood why DNCL1 is encapsulated into the RABV viral particles. In contrast, G interaction with MTs appears to be important in mediating the viral particles transport within neurons and thus RABV spreading throughout the CNS, but additional studies are needed to experimentally substantiate this notion.

| Adenovirus
ADV infects numerous species of vertebrates, including humans, causing diverse pathologies, mainly in respiratory and ocular systems. In the last decades, ADV has become a valuable potential therapeutic vector for gene therapy and vaccine design. 90 ADV is a group of linear double-stranded DNA (dsDNA) viruses with a 26-48 kbp genome that associates with multiple DNA-binding core proteins and is packed in a complex icosahedral capsid made by the coat proteins ( Figure 3B). 91 The genome of ADV also encodes auxiliary proteins, such as the protease, which play key roles during the viral life cycle. 92 After interacting with the cell surface receptor, ADV enters the host cell through endocytosis and the lytic activity of the viral protein VI promotes ADV escape from endosomes. 93 Once it is delivered in the cytoplasm, the viral DNA and associated core proteins are transported toward the nucleus, 94 where the viral genome is imported for transcription and replication. 95 Assembly and coating of new virions occur near the replication sites in the nucleus, while egression takes place through still unclear lytic and/or nonlytic pathways. 90 Electron microscopy analyses in 1973 already reported a possible interaction between ADV viral particles and MTs, suggesting that this cellular scaffold may mediate the transport of ADV virions within the host cell cytoplasm. 96 Similar to the early RABV experiments, the yeast two-hybrid system has allowed to uncover that the early tran- During ADV cell entry, integrins, which serve as virus co-receptors, are activated and trigger the downstream activation of protein kinase A (PKA). 102 PKA phosphorylates DYNC1LI1 and this modification promotes DYNC1LI1 binding to hexon. 103 Interestingly, the interaction between hexon and dynein is pH-dependent. 55 In particular, a low pH induces reversible and subtle conformational changes in the hypervariable region 1 of hexon, which is necessary for its binding to dynein. 104 Such low pH can be encountered by ADV in endosomes during cell entry, although it has been shown that is not a requirement for an efficient viral entry. 105 In endosomes, the hexon proteins are primed for their interaction with DYNC1LI1 and when the virions reach the cytoplasm, the binding to DYNC1LI1 permits their efficient transport into the nucleus. 104 Accordingly, when the hexon-DYNC1LI1 interaction is prevented, the transport of the ADV viral particles into the nucleus is completely inhibited. 104 Thus, while the hexon-DYNC1LI1 interaction mediates the transport of ADV virions that have entered the cell into the nucleus, the eventual importance for ADV life cycle of the binding of E3-14.7K, protease and possibly E1A with the dynein subunits remains to be understood.

| VIRAL HIJACKING OF KINESINS
The kinesin family has a large number of members and functions and this is probably increasing the diversity of the roles that they may

| Herpes simplex virus
HSV can cause severe diseases in different human organs and its clinical relevance mainly relies on its ability to establish latency in neurons, resulting in lifelong infections. 109 HSV contains a 150 kbp, linear dsDNA genome, which is packed in an icosahedral capsid. The capsid is enclosed by proteins forming the tegument, which in turn is surrounded by the viral envelope ( Figure 3C). 110 After binding to the cell surface, the HSV envelope fuses with the host plasma membrane, delivering the unenveloped virus, that is, the tegumented capsid, into the cytoplasm. 111 After losing the tegument, the capsid is transported into the nucleus, 112 where transcription and replication take place. 113 The newly assembled capsids are transported from the nucleus into the cytoplasm, in which they acquire the tegument and the envelope at the Golgi, before being transported toward the cell periphery for exit and spread. 114,115 The first direct observation of a specialized MT-mediated anterograde transport of HSV was obtained using a cellular system in which human neuronal and epidermal tissue were co-cultured. It was reported that the axonal transport of HSV toward epidermal cells takes place adjacently to MTs. 116 This observation was later confirmed with studies in both human and rat neurons, which showed that this MT-mediated anterograde transport indeed depends on MTs and that the cargo is unenveloped HSV viral particles. 117 The directionality of this transport suggested that kinesins might mediate the interaction between HSV and MTs. In vitro and in vivo analyses indeed proved that kinesin-1 can directly interact with the HSV capsid (Table 1), mainly through tegument proteins. 56 It was subsequently revealed that in human neurons, kinesin-1 heavy chain KIF5B interacts via the binding site at the amino acids 867-894 with the HSV tegument protein US11 (Table 1), leading to the conclusion that US11 has a major role in the MT-mediated anterograde axonal transport of unenveloped HSV. 62 UL36 is the largest tegument protein and studies on a HSV strain lacking this protein showed a decrease in the anterograde transport of unenveloped HSV, suggesting that UL36 or one of its binding partners may also participate in this movement. 118 In fact, the formation of the complex between UL36 and UL37, another tegument protein, is required for unenveloped HSV MT-mediated anterograde transport. 119 Very recently, colocalization and co-immunoprecipitation experiments have uncovered an interaction between UL36 and kinesin-1 (Table 1). 57 Interestingly, the authors of this work showed that the binding to UL36 is responsible, at least partly, for the packaging of kinesin-1 into the tegument and making it structural component of HSV. This incorporation allows HSV to carry kinesin-1 between cells and use it intracellularly, upon cell entry. 57 In particular, the interaction between UL36 and kinesin-1 is essential for the nuclear import of unenveloped HSV, because it mediates the anterograde transport from the MTOC close to the nuclear envelope surface. 57 Of note, kinesin-2 was also shown to be associated with the HSV capsid (Table 1), although it is unclear whether the kinesin-2-capsid interaction is direct or indirect and whether this interaction has a role in HSV life cycle. 56 After transcription and replication, the newly synthesized HSV capsids exit the nucleus through a three-step process: the capsids bud within the inner nuclear membrane, accumulate in the perinuclear space and they are finally released in the cytoplasm as membrane-less capsids. 120 Subsequently, the HSV capsids bud into cytoplasmic organelles that appear to be part of the trans-Golgi and/or endocytic systems. 121 To investigate the MT-dependent transport of HSV capsids in vitro, a study isolated these cytoplasmic HSV-containing organelles from cells infected with a HSV strain expressing GFP-tagged VP26, an HSV capsid subunit. 122 The results provided evidence that kinesin-dependent transport along MTs directs trafficking of HSVcontaining organelles, leading to the suggestion that this suggesting may play role in HSV egression. 122 The glycoproteins that are part of the HSV envelope follow a pathway different than the one of RÍO-BERG ultimately form the capsids. 117 In particular, the HSV UL56 tegument protein associates with the kinesin-3 protein KIF1A via a domain within its residues 69-217 (Table 1). 66 As KIF1A mediates anterograde axonal transport of synaptic vesicles, the model that has been proposed is that UL56 hijacks this trafficking route to transport the vesicles containing the viral glycoproteins to the Golgi, where HSV capsid enveloping takes place. 66 MT-mediated anterograde transport plays a role in HSVcontaining organelles both before and after the viral envelopment.
Aiming to clarify which kinesins take part in these steps, the function of several of them was assessed in neuronal cells by fluorescent microscopy. 58 It was found that kinesin-1 proteins KIF5A, KIF5B and KIF5C, play a major role in promoting MT-mediated anterograde axonal transport of complete HSV virions (Table 1). 58 The UL36 and UL37 tegument proteins, however, are not required for this movement. 123 In contrast, results in murine retinal ganglion cells indicated that the envelope protein US9 is implicated in MT-mediated anterograde axonal transport of enveloped HSV. 124 Through in vitro pulldown experiments, it was revealed that US9 interacts with KIF5B (Table 1). 63 Glycoproteins gI and gE have also been shown to be involved in the MT-mediated anterograde transport of enveloped HSV. 125 A subsequent study connected these two apparently discrepant observations by showing that these three proteins form a complex, in which gE binds simultaneously to both gI and US9. 126 The gI-gE-US9 complex, however, does not directly modulate the MTmediated anterograde transport. 127 Instead, in association with other tegument proteins (i.e., VP22, UL11, UL16 and UL21), this complex regulates the envelopment step, which is key in inducing the correct axonal transport of enveloped HSV. 127,128 Thus, it has been postulated that there may be one or more proteins other than gI, gE and US9, which interact with kinesins to mediate the anterograde transport of complete HSV particles via MTs.
All in all, there are still some mechanistic gaps to have a complete picture of how HSV uses kinesins for the MT-mediated anterograde transport of its components, to promote both virion assembly and egression. What the available data indicate is that there are redundant interactions between HSV proteins and kinesins. 56,62 This notion is indirectly supported by the fact that most of the mentioned HSV proteins influence viral pathogenicity, but they are not essential for replication. [129][130][131] Because MT-mediated anterograde transport is such an important step in HSV spreading, it is also conceivable that several redundant mechanisms exist, making it difficult to address this question experimentally.

| Human immunodeficiency virus 1
HIV-1 infects cells of the immune system, principally T cells and macrophages, causing the acquired immunodeficiency syndrome (AIDS).
HIV-1 consists of two copies of RNA that by associating with the nucleocapsid protein, the viral protease, the reverse transcriptase and the integrase, form the reverse transcription complexes (RTCs). The RTCs are packaged into a proteinaceous capsid, which in turn is surrounded by the matrix and the viral envelope ( Figure 3D). 132 The glycoproteins that are anchored in the envelope bind to the host cell receptor at the plasma membrane to initiate the HIV-1 cell entry through membrane fusion, 133 resulting in the cytoplasmic release of the viral capsid. 134 It was recently shown that cytoplasmic HIV-1 capsids are subsequently imported into the nucleus via nuclear pore complexes. 135 In the nucleus, the reverse transcriptase transcribes the viral linear ssRNA into DNA, 136 142,143 To find the effector proteins of this process, a study knocked down KIF5B, 144 which was known to be important for the uncoating of ADV that precedes the nuclear import of its genome. 145 It was found that this kinesin, in association with MTs, plays an important role in HIV-1 nuclear import and capsid uncoating. 64,144 KIF5B induces the re-localization of the nuclear pore complex component

Nup358 into the cytoplasm during HIV-1 infection, allowing Nup358
to directly interact with the capsid (Table 1). 64 The cooperation of Nup358 and KIF5B leads the transport event that results in the HIV-1 genome nuclear import. 64 Later, it was revealed that KIF5, through both KIF5A and KIF5B, binds the hexamers formed by the HIV-1 capsid protein via the fasciculation and elongation factor zeta-1 (FEZ1; The assembly of new HIV-1 virions takes place at the plasma membrane of T cells and this implies that the newly synthesized proteins have to reach this location. 147 MTs are involved in the vesicular trafficking to the plasma membrane of Gag and Env, two polyprotein precursors that are required for the correct assembly and spread of HIV-1. 148 A yeast two-hybrid study found that kinesin KIF4A binds Gag (Table 1), an interaction that was confirmed in infected mammalian cells using pull-down experiments. 67 Subsequently, it has been revealed that the KIF4A-Gag interaction regulates Gag intracellular trafficking to the plasma membrane, but also Gag stability. 149 A different study showed an additional function for KIF4A in the early steps of HIV-1 infection. In particular, KIF4A also binds the HIV-1 matrix protein (Table 1) and promotes its association with the MT endbinding protein EB1. 68 The activation of EB1 recruits other proteins that induce the modifications to MTs, such as acetylation. This results in HIV-1 enhancing MT stability, which increases the MT-dependent virus cell entry and ultimately the viral infection. 68 The ability of viruses to stabilize MTs through acetylation was already documented previously for ADV 150 and HSV. 151 KIF3A, which was initially identified in a siRNA-based genomewide screen, 65 plays an important and specific role in the last steps of the life cycle of HIV-1 in macrophages, one of the cell types targeted by this virus. In these cells, HIV-1 assembly takes place in virus containing compartments, which are derived from endosomes. 147 It has been revealed that KIF3A is anchored to the membrane of these compartments (Table 1) 154 However, no direct interactions between the virus and myosin motors take place in this process. 153 In this paragraph, we will discuss the interactions of IAV virus and HPV with myosins. However, viruses such as Marburg virus, 155 vaccinia virus, 156 human rhinovirus 157 and human cytomegalovirus 158 have also been shown to hijack myosins during their viral life cycle.

| Influenza A virus
IAV is responsible for most of the seasonal epidemics of influenza, by infecting the human respiratory epithelium and causing pathologies that range from mild respiratory diseases to death. 159 The IAV genome is 13.5 kb and consists of eight segments of single-stranded RNA (ssRNA) that are encapsulated in a capsid formed by the matrix proteins M1. This is further surrounded by an outer lipid envelope, where the matrix M2 ion channels traverse ( Figure 3E). 160 The haemagglutinin (HA) proteins on the virion surface bind to the host cell receptors, triggering the IAV cell entry through endocytosis. 161 The IAV-containing endosomes are transported towards the perinuclear area and during this transport, their pH lowers progressively. The acidification of endosomes triggers the fusion of the IAV viral particles with the limiting membrane of this organelle and the cytoplasmic release of the capsid. 162 The acidic environment also allows the subsequent dissociation of vRNPs from the M1 proteins and between each other. 163 Each vRNP consists of a ssRNA segment, the nucleoprotein (NP) and the RNA-dependent RNA polymerase complex formed by PA, PB1 and PB2. Cytoplasmic vRNPs are then imported into the nucleus, in which ssRNA transcription and replication take place. 164 Eventually, the newly synthesized IAV particles are assembled at the plasma membrane and released through outward budding. 160 The visualization of individual IAV during infection in mammal cells using live-cell fluorescence microscopy shed light on the IAV entry steps. It was described that right after entry and before MTdependent transport, IAV-containing endosomes colocalize with Factin in the cell periphery and undergo a long stage of slow and directed movement, which indicates that they could be re-localized by myosin motors. 165,166 Because myosin VI had been proved as the retrograde motor of endocytic vesicles along with F-actin in cell periphery, 167  IAV infection and that myosin VI plays an important role. 169 However, these elements are dispensable in non-polarized cells. 169 The role of myosin VI in IAV entry in polarized cells was further confirmed by live-cell fluorescent microscopy and showed that myosin VI is responsible for the transport of the IAV-containing endosomes along F-actin (Table 1). 77 Moreover, it was observed that when the switch to dynein-directed transport takes place at the actin-MTs intersections, myosin VI remains attached to endosomes. 77 IAV also enters host cells through macropinocytosis and myosin appears to be differently involved in this alternative infection mechanism. 76 By screening a library of inhibitors and using a negative dominant of myosin II, it was revealed that myosin II light chain activation through phosphorylation by the myosin light chain kinase MLCK, is essential for IAV cell entry through micropinocytosis (Table 1). 76 This role of myosin II light chain phosphorylation in IAV replication was supported by another study that used genetic and chemical inhibition of several pathways. 170 In particular, it was concluded that the signalling cascades downstream of the RhoA/Rho-kinase, phospholipase C/protein kinase C and HRas/Raf/MEK/ERK converge in the RÍO-BERG E ET AL.
phosphorylation of the myosin light chain and induce the remodelling of the actin cytoskeleton. Real-time PCR analysis confirmed that this function is needed to promote IAV proliferation. 170 The cytoplasmic release of the vRNPs takes place near the nucleus, where endosome transport is directed via MTs. 171 A study using RNA interference and inhibitors of actin polymerization and myosin II, found that the actin cytoskeleton and myosin II are also required in this event. 72 Specifically, histone deacetylase 6 (HDAC6) acts as a bridge between the viral M1 and myosin II (Table 1), which, together with dynein and MTs, generate opposing physical forces that promote the disassembly of the M1 capsid and cytoplasmic release of vRNPs during the fusion of the IAV viral particles with the limiting membrane of endosomes. 72 As previously mentioned, physical forces also initiate the HIV-1 capsid disassembly, 138 and mediate other steps in the infection of several viruses as comprehensively reviewed by Liu et al. 172 The actin cytoskeleton is also needed for the distribution of newly synthesized M1, HA and NP proteins at the plasma membrane, 173 and myosin II is at least directly implicated in HA dynamics (Table 1). 73 The interactions of the actin cytoskeleton and IAV structural proteins, 174 and binding of M1, vRNPs and HA with F-actin, [175][176][177][178] eventually promote the correct assembly and budding of new IAV virions. 73,173 Actin filaments participate in the formation of structures named tunnelling nanotubes, which are a physical connection between cells that are used for the exchange of organelles and other material. 179 Interestingly, using immunofluorescence and live-cell microscopy, it has been shown that vRNPs but not the nonstructural protein 1 (NS1), can spread from one cell to another through the tunnelling nanotubes, as a mode of propagation alternative to the common cell egress and entry mechanism. To move through these intercellular connections, vRNPs depend on F-actin and it is has been suggested that they may employ a myosin motor. 180 Because myosin Va (Table 1) is present in the tunnelling nanotubes, 179 this motor protein could be a putative candidate for mediating vRNPs transport through these structures, but this remains to be experimentally explored.
To date, the actin cytoskeleton and myosin motors have been shown to be involved in multiple steps of the IAV life cycle, although the identity of the myosins that specifically acts in most of those steps remains unknown. The characterization of the binding mechanism between IAV proteins and myosins, which probably require cargo adaptors, is also still lacking.

| Human papillomavirus
HPV infect skin and mucosa epithelial cells, in which they induce benign tumours that can progress to malign. 181 Figure 3F). Apart from the structural proteins, the genome encodes for the E1 and E2 proteins, which are responsible for HPV replication and E4, E5, E6 and E7 proteins, which play diverse roles in HPV life cycle. 183 188 To find the mechanism responsible for filipodia induction, cells were infected with HPV31 while exposed to a panel of various inhibitors. This approach revealed that tyrosine kinases, phosphoinositide 3-kinases and myosin II are involved in the actin rearrangements that are needed for the filopodia formation and HPV31 cell entry (Table 1). 74 The authors hypothesized that HPV31 activates signalling pathways centred around these kinases to promote filipodia formation and myosin II-driven viral particle transport toward the cell body to ensure infection. 74 Single-particle tracking of HPV16 by live-cell fluorescence microscopy showed the role of F-actin in mediating the retrograde transport from filopodia towards the cell body in HeLa and keratinocyte cell lines. 189 The contribution of myosin was verified by inhibiting myosin II and MLCK with blebbistatin and ML-7, respectively, which resulted in an almost instantaneous stop of the viral particle transport via filopodia. 189 However, neither actin cytoskeleton nor myosin II played a role in HPV16 infection in confluent cells, indicating that these components are rather required for filopodiamediated entry and not intracellular movement. A different mechanism for HPV16 cell entry, characterized by the formation of small invaginations at the plasma membrane and therefore sharing some similarities with micropinocytosis, has also been described. 75 This cell entry mechanism also involves the actin cytoskeleton, MLCK and myosin II ( This interaction was also confirmed by immunoprecipitation experiments, which also uncovered that HPV18 E6 triggers the polyubiquitination and subsequent proteasomal degradation of GIPC1. 193 Interestingly, GIPC1 is a known adaptor for myosin VI that promotes the movement along actin filaments of its binding partners such as GLUT1. 194 Although it remains to be established which are the effects of this interaction on HPV infection, these findings suggest a strategy of HPV18 in blocking myosin VI-mediated anterograde transport. A study examining myosin 1b (Table 1) expression in cervical cancer found that myosin 1b is overexpressed in HPV-positive cervical cancer cells and this results in increased proliferation, migration and invasion. 69 Because HPV16 E6 and E7 proteins are the main responsible factors for the initiation and progression of cervical cancer, 195 they were knocked down in HPV infected cells to determine whether they are involved in the upregulation of myosin 1b. 69 The individual depletion of E6 or E7 indeed reduced the myosin 1b protein levels, suggesting that these viral proteins are responsible for the myosin 1b overexpression in cervical cancer cells. 69 Pull-down and coimmunoprecipitation assays have been employed to identify binding partners of HPV8 E6 and E7 proteins and one of them is myosin 1c (Table 1). 70 Myosin 1c is the nuclear isoform of myosin 1 and is an essential component of the cellular RNA polymerase I transcription complex. 196 Importantly, the interaction of HPV8 E7 with myosin 1c leads to a decrease in the activity of the RNA polymerase I, downregulating the expression of preribosomal RNA. 70 The authors of this study proposed that HPV8 may use this mechanism to interfere with cellular transcription to promote a more efficient replication of its own viral genome. Myosin 1c has also been shown to interact with E2 proteins (Table 1) from different HPV, that is, HPV18 and HPV5, in immunoprecipitation assays. 71 The E2-myosin 1c complex is also associated with the B-WICH complex, 71 which mediates RNA polymerase I and III transcriptions through chromatin remodelling. 197 Because the siRNA downregulation of myosin 1c increases the E2-dependent replication of the HPV5 genome, it has been suggested that myosin 1c represses HPV5 replication through the recruitment of E2 to the B-WICH complex. 71 Overall, the role of myosins in HPV life cycle appears to be restricted to its cell entry and the regulation of transcription in the nucleus. Of note, however, other steps of HPV life cycle may require MTs.

| DISCUSSION
The cytoskeleton and its associated motor proteins are often coopted by viruses to promote their own intracellular replication and distribution within the infected tissues. In accordance with the directionality provided by the cytoskeleton, in particular MTs, it is important to note that the viral life cycle often starts and finishes in the plus end. Given that anterograde or retrograde transport occurs, binding to the correct motor protein(s) allows to take the right track to promote a determined step of the viral life cycle. Despite the evidence pointing out that a large number of viruses belonging to different viral families interact with motor proteins, how they hijack dyneins, kinesins and myosins remains largely enigmatic. The analysis of the subversion strategies adopted by different viruses and presented in this review, demonstrates poor mechanistic commonalities and it has led us to conclude that three aspects that must be considered in an investigation on the interaction between a virus and motor proteins.
First, a full understanding of the viral life cycle is necessary to eventually predict and investigate which steps may require the cytoskeleton.
This can be exemplified by HPV, as the mechanism used for cell egression makes evident that there is probably no cytoskeleton role. 181 Second, knowledge of the distribution of the viral proteins, within Overall, most of the initial observations about a possible role of the cytoskeleton in the life cycle of viruses have been done through microscopy approaches and using compounds that disrupt a specific cytoskeletal scaffold such as colchicine or nocodazole for MTs and cytochalasin D for F-actin. 80,169,187 Although these approaches are effective to show a possible implication of a specific cytoskeletal scaffold in a certain viral life cycle, the observed defects could also be because of the disruption of one or more processes affecting RÍO-BERG indirectly the pathway that is directly hijacked by the virus. As a result, the most important but also challenging part of an investigation is unveiling the precise molecular mechanism underlying the interaction between the virus and the studied cytoskeletal elements. That is, which viral protein(s) is (are) involved in the subversion and which cytoskeleton-associated protein(s) or motor protein(s) is (are) hijacked.
When searching for these proteins, an important consideration is the identification of the adaptors, which confer the cargo-specificity to motor proteins. Because each virus connects a different cargo with the cytoskeleton, it is probable that in most of the situations, the motor protein-adaptor system is relevant for the viral hijacking mechanism, although this has to be established in each situation. The examples we chose in this review, have been described in some molecular detail, yet there are still considerable conceptual gaps between the described interaction and the corresponding function. In

CONFLICT OF INTEREST
The authors have declared no conflicts of interest for this article.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1111/tra.12835.