Vaccinia virus proteins A36 and F12/E2 show strong preferences for different kinesin light chain isoforms

Vaccinia virus (VACV) utilizes microtubule‐mediated trafficking at several stages of its life cycle, of which virus egress is the most intensely studied. During egress VACV proteins A36, F12 and E2 are involved in kinesin‐1 interactions; however, the roles of these proteins remain poorly understood. A36 forms a direct link between virions and kinesin‐1, yet in its absence VACV egress still occurs on microtubules. During a co‐immunoprecipitation screen to seek an alternative link between virions and kinesin, A36 was found to bind isoform KLC1 rather than KLC2. The F12/E2 complex associates preferentially with the C‐terminal tail of KLC2, to a region that overlaps the binding site of cellular 14‐3‐3 proteins. F12/E2 displaces 14‐3‐3 from KLC and, unlike 14‐3‐3, does not require phosphorylation of KLC for its binding. The region determining the KLC1 specificity of A36 was mapped to the KLC N‐terminal heptad repeat region that is responsible for its association with kinesin heavy chain. Despite these differing binding properties F12/E2 can co‐operatively enhance A36 association with KLC, particularly when using a KLC1‐KLC2 chimaera that resembles several KLC1 spliceforms and can bind A36 and F12/E2 efficiently. This is the first example of a pathogen encoding multiple proteins that co‐operatively associate with kinesin‐1.

resulting in a reduced plaque size in cell culture and attenuated virulence in vivo. [13][14][15][16][17][18] The movement of IEVs from the site of wrapping to the cell surface is incompletely understood but is mediated by the kinesin-1 MTassociated motor complex. 11 Kinesin-1, also known as conventional kinesin, is the prototype member of the kinesin protein superfamily. 19 It consists of a dimer of kinesin heavy chain (KHC) molecules that have 3 isoforms encoded in mammals by the KIF5A, KIF5B and KIF5C genes.
Each KHC consists of an N-terminal MT-binding ATPase motor domain, an extensive coiled-coil dimerization domain and a C-terminal cargo interaction domain (see Figure 1A for a diagrammatic representation). While some kinesin-1 cargos, such as the mitochondrial associated MIRO-MILTON complex, interact directly with the KHC C terminus, 20 many require the presence of 2 copies of the kinesin-light chain (KLC) adaptor protein. KLCs consist of an N-terminal coiled-coil domain responsible for dimerization and KHC interaction, 6 tetratricopeptide repeat (TPR) motifs that each form a helix-turn-helix structure, which stack to form a stable protein interaction domain, and a flexible C-terminal tail ( Figure 1A). In mammals 4 isoforms have been identified, each encoded by a separate gene. Both KLC1 and KLC2 are expressed ubiquitously, though KLC1 is often described as being enriched in neuronal cells, 21 KLC3 is limited to spermatid cells 22 and KLC4 expression remains to be fully characterized.
At least 6 VACV proteins are directly associated with the wrapping membranes that form the 2 outer envelopes of IEVs ( Figure 1B). These include the transmembrane proteins A56, 23 B5, 18,24,25 A34, 15,26,27 A33 15,26,27 and A36, 16 and the palmitoylated protein F13. 14 A36 is associated exclusively with the outer of the 2 IEV envelopes and, upon fusion of this envelope with the plasma membrane, accumulates at the site of CEV attachment. 28 Here A36 in complex FIGURE 1 Testing the interaction of VACV IEV proteins with kinesin-1 by co-IP with epitope-tagged KLC1 or KLC2. A, Schematic diagram of the kinesin-1 complex that mediates trafficking of cargos along microtubules (MT) from the slow-growing-end to the more dynamic + end oriented towards the cell periphery. Kinesin-1 is usually represented as a heterotetramer consisting of 2 copies of KHC and 2 copies of KLC. KHC proteins (~110 to 130 kDa) possess an N-terminal ATPase MT-binding motor domain (shown in purple), a long coiled-coil binding domain and C-terminal tail domain of unknown structure. KLC proteins (51 to 76 kDa depending on the isoform) consist of a short N-terminal coiledcoil region responsible for binding to KHC, an α-helix rich structural region consisting of 6 TPR motifs and a C-terminal tail. B, The VACV IEVassociated proteins. During infection some IMV are transported from virus factories on MTs and are wrapped by cellular membranes containing several VACV transmembrane and acylated proteins to form IEV particles. IEVs associate with the kinesin-1 complex and are transported to the cell surface where F12/E2 dissociate. Virions are externalized by exocytosis and either remain bound to the cell surface as CEVs or are released as EEVs. CEVs can induce the polymerization of actin beneath the CEV particle and this requires the A36 protein. C, Co-precipitation of IEV proteins with epitope-tagged KLC1 or KLC2. FLAG-tagged KLC1, KLC2 and GFP were expressed in HEK293T cells by plasmid transfection. Cells were infected 24 h later with VACV at 5 pfu/cell. Clarified cell lysates were generated 12 h post-infection (hpi) and used to immunoprecipitate the FLAG-tagged proteins. The immunoprecipitates were analysed by SDS-PAGE and immunoblotting. Blots shown are representative of several experiments (n = 3) using either vF12-HA or vE2-HA.
with A33 triggers formation of actin tails. 12,29,30 The A36/A33 complex is also expressed at the cell surface prior to new virions being made and can induce actin tails beneath superinfecting virions repelling them to enhance virus spread. 31 A36 is the only protein described to interact directly with kinesin-1 and link it to IEVs. 32 A36 possesses a bipartite kinesin-interaction motif consisting of a tryptophan residue surrounded by 1, 2, or 3 acidic residues (referred to as a WE/D motif ), shared by many cellular kinesin-1 interacting proteins. 33,34 The structure of a KLC2 TPR domain co-crystallized with a WE/D motif-containing peptide (derived from the cellular SifAkinesin interacting protein, SKIP) showed binding of the WE/D motif into a groove formed by the second and third TPR motifs on the inner surface of the KLC TPR domain. 35 A36 probably interacts with KLC in the same manner through its own WE/D motifs.
F12 is a 65-kDa cytoplasmic protein that, like A36, is associated with IEVs but not IMVs, EEVs or CEVs, 36 39 The F12/E2 complex associates with kinesin-1 through an interaction of E2 with the C-terminal tail of KLC2. 40 In the absence of A36 IEVs still undergo MT-mediated egress. 41 The ability of F12/ E2 to interact with KLC2 may explain how IEVs lacking A36 can move in a MT-dependent manner, however F12/E2 has not been shown to link kinesin-1 to IEVs. Alternatively, other link(s) between IEVs and the motor complex mediated by viral or cellular proteins may exist.
In this report, all known VACV-encoded IEV-associated proteins (that are absent from IMV particles) were screened by co-

| RESULTS
A yeast-2-hybrid screen of the cytoplasmic portions of some of the VACV IEV proteins identified A36 as the only link between IEVs and kinesin-1. 32 However, only the TPR region of KLC, the region often associated with cargo interaction, 42 was tested in this study. A later report that the F12/E2 complex interacted with the C-terminal tail of KLC 40 showed that other parts of the KLC protein are also important for cargo interaction and other VACV proteins are involved. Therefore, to test if other IEV proteins interact with KLC, all IEV proteins involved in the formation and egress of IEVs, expressed at endogenous levels during infection, were re-screened by coimmunoprecipitation with full-length KLC.
FLAG-tagged KLC1 and KLC2 were expressed in HEK-293 T cells that were infected subsequently with VACV. The FLAG-tagged KLC was immunoprecipitated from clarified cell lysate and immunoblotted to determine if any of the IEV proteins co-precipitated ( Figure 1C).
Antibodies were available for A33, A34, B5, F13 and A36, and so to detect F12 or E2, cells were infected with vF12-HA (a recombinant VACV expressing HA-tagged F12 36 ) or vE2-HA (expressing HAtagged E2 40 ). The results show that A33, A34 and B5 were not coprecipitated with either KLC. This is consistent with the fact that the majority of these proteins are within the luminal space of the IEV envelope rather than being cytosolic ( Figure 1B). In contrast, F12 and E2 were both co-precipitated with KLC2, confirming previous observations that the F12/E2 complex associates preferentially with the KLC2 isoform. 40 Optimization of experimental conditions showed that while both F12 and E2 co-precipitate more efficiently with KLC2, they also co-precipitate with KLC1 to a lesser degree.
A36 was shown to interact with KLC1 by yeast-2-hybrid, 32 with KLC2 by FRET (Förster resonance energy transfer) microscopy 43 and both KLC1 and KLC2 by co-precipitation when overexpressed ectopically. 33,40 Therefore, it was surprising to find that A36 showed a very strong association with KLC1 and practically no binding to KLC2 ( Figure 1C). In many of the previous experiments, A36 was not only overexpressed ectopically but was often expressed in a soluble form lacking its transmembrane domain. The results in Figure 1C represent the first time this interaction has been shown using full-length A36 expressed at endogenous levels during virus infection.
In addition to A36 and the F12/E2 complex, F13 also coprecipitated with KLC. Unlike A36, F12 and E2, F13 did not show a KLC isoform specificity, and the levels of F13 co-precipitating with KLC varied considerably between experiments. To confirm the interaction, the reciprocal co-immunoprecipitation was attempted with the anti-F13 antibody, but without success. As an alternative approach, the reciprocal co-precipitation was then attempted by cotransfecting HEK-293 T cells with plasmids expressing HA-tagged F13 (using a codon optimized allele) along with FLAG-tagged KLC.
Immunoprecipitation of ectopically-expressed HA-F13 coprecipitated FLAG-KLC ( Figure 2). However, the efficiency of coprecipitation was highly variable and appeared to be sensitive to small differences in detergent concentration. F13 is an abundant IEV protein present between the IMV and IEV inner membrane and on the cytosolic face of the IEV outer membrane ( Figure 1B) but is also present in other cellular membranes. F13 does not possess a transmembrane domain and instead associates with membranes via palmitoylation of cysteines 185 and 186. 44 Mutation of these residues to serine alters F13 localization and abrogates wrapping of IMV to form IEV. Introducing these same mutations into the codon-optimized HAtagged F13 and repeating the co-precipitation resulted in a loss of KLC co-precipitation ( Figure 2). While this does not discount completely that F13 is a kinesin-1-interacting protein (palmitoylation may be required for correct protein folding), taken together with the sensitivity of the co-precipitation to detergent, these results suggest that the observed F13/KLC interaction is indirect or an artefact of the experimental conditions.

| F12/E2 displaces the cellular KLC interacting 14-3-3 protein from KLC2
The region of KLC2 responsible for the enhanced association of F12/E2 has been mapped using KLC1/KLC2 chimaeras to the Cterminal tail of KLC2. 40 To map this interaction more accurately, a series of mutants of KLC2 were made lacking the C-terminal 16 (KLC2ΔC16), 46 (KLC2ΔC46) or 88 (KLC2ΔC88) amino acids ( Figure 3B). Precipitation of these mutants from cells infected with vF12-HA showed that only the last 16 amino acids were dispensable for F12/E2 binding and further truncation resulted in F12 coprecipitation levels equivalent to those with KLC1 ( Figure 3D).
To our knowledge, the only other proteins that interact with the C-terminal tail of KLC2 are the family of cellular 14-3-3 proteins 45 that function as scaffolds involved in the assembly and subcellular localization of signalling complexes. 46 Binding of 14-3-3 to KLC2 is dependent on the phosphorylation of serines 542 and 579 47 and these serines are present in the region essential for F12/E2 binding. To test if these residues were needed for F12/E2 interaction, FLAG-KLC2 with serines 542 and 579 mutated to alanines, either individually or together (as shown in Figure 3C) were immunoprecipitated from vF12-HA-infected cells. As expected, the mutated KLC2 showed a reduction or loss of 14-3-3 co-precipitation ( Figure 3E). However, F12 co-precipitation was not affected, indicating that F12/E2 association with KLC2 is independent of 14-3-3 interaction and does not require serines 542 or 579 to be phosphorylated. Additional evidence that F12/E2 and 14-3-3 share overlapping binding sites but do not bind co-operatively to KLC was provided by the ability of E2 to interfere with the KLC2/14-3-3 interaction. In uninfected cells 14-3-3 is coprecipitated with ectopically-expressed FLAG-KLC2 (Figure 3 F).
When E2 was expressed ectopically in these cells E2 co-precipitated with FLAG-KLC2. Expression levels of 14-3-3 are not altered in the presence of E2, but it no longer co-precipitates with FLAG-KLC2.

| Mapping the interaction of A36 to KLC1
KLC1 and KLC2 share a high degree of amino acid similarity, particularly within TPRs 1, 2 and 3 ( Figure 4A) were tested for interaction with A36 but neither mutation altered KLC interaction with A36 ( Figure 4Dii).
This indicated that the specificity for the A36 KLC interaction was upstream of the TPRs. To pinpoint the exact region responsible, additional chimaeric KLCs were generated in which parts of the Nterminal region were switched between KLC1 and KLC2 ( Figure 5A).
Only chimaeras possessing the heptad repeat region from KLC1 were able to bind A36 and switching any other region of KLC did not affect A36/KLC co-precipitation ( Figure 5B). As expected, all of the chimaeric KLCs retained their ability to bind to KHC. Subsequently, KLC1 truncations were generated containing either the N-terminal HR region (KLC1 HR) or TPR domain with the C-terminal tail region (KLC1 TPR + CT) and tested for their ability to bind A36. Only KLC1 TPR + CT was able to interact with A36, while the HR region on its own was unable to bind A36 ( Figure 5C).    expressing F12 inducibly (T-REx 293-F12-HA) and that was transfected with FLAG-KLC1 and subsequently infected with vΔF12. The levels of A36 that co-precipitated with KLC1 were higher when F12 was present ( Figure 6A). Quantification of repeated experiments confirmed this F12 enhancement of A36 co-precipitation with KLC1

|
( Figure 6Bi). Doxycycline treatment of cells did not alter the A36 and F12/E2, so that all components may be present in the same complex, showed reduced A36-KLC interaction in the absence of F12 or E2 ( Figure 7B). This suggests that there is co-operativity in the association of the different components of the IEV trafficking complex, especially if all components are able to efficiently associate.

| DISCUSSION
VACV proteins A36, F12 and E2 all associate with kinesin-1 during IEV egress and influence IEV egress efficiency. Of these proteins only A36 is associated directly with the IEV particle via a transmembrane domain, yet deletion of A36 is less detrimental to virus egress than deletion of either F12 or E2, and IEVs lacking A36 are still transported in a MT-dependent manner. 41 The interaction of A36 with KLC was found initially using a yeast-2-hybrid screen that tested the cytoplasmic portions of several IEV proteins (A33, A34, A36, B5 and F12) for binding to the KLC TPR domain. 32 Here another screen was undertaken using full-length KLCs and the more physiologically relevant condition of virus-infected cells where full-length IEV proteins were expressed at natural levels ( Figure 1). This co-precipitation screen identified A36 and F12/E2 as expected, but also F13. However, the F13 interaction was variable and the observation that a mutant F13 that was unable to associate with membranes was unable to interact with KLC suggested that the F13/kinesin-1 interaction was either non-specific or indirect, perhaps mediated by another protein, possibly of cellular origin.
The co-precipitation screen confirmed the previous report that F12/E2 bound KLC2 preferentially 40 and mapped the interaction to a region that overlaps the region required for association with the cellular 14-3-3 scaffold protein. 47  The processes that convert ATP hydrolysis by the kinesin motor into processive motion along MTs has been elucidated in detail. 48 How the activity of the kinesin complex is regulated and how it interacts with its cargos is less well understood. The regulation of kinesin activity is essential. In its absence, kinesin-1 motor complexes would be constitutively active resulting in rapid depletion of cellular ATP, and kinesin-1 redistribution away from where it is needed. Kinesin-1 regulation relies heavily on autoinhibition. 49  was over-expressed ( Figure 6). Using a chimeric KLC1/2 molecule that bound A36 and F12/E2, a reduction in A36 binding to KLC was observed in the absence of F12 or E2 (Figure 7). Over 16 different KLC1 spliceforms have been described in humans and mice, each with identical heptad repeat and TPR domains but differing in the length and makeup of their C-terminal tail. 56 Some of these alternative C-terminal tails share amino acid sequence similarity with KLC2.
The cellular 14-3-3 protein, whose KLC interaction overlaps that of F12/E2, can associate with KLC1 spliceform J. 45 It is therefore conceivable that a KLC1 spliceform exists with which F12/E2 and A36 can bind co-operatively and may be the actual target of VACV kinesin-1 engagement.
The interaction of cargo with KLC is itself regulated by autoinhibition. 57 A negatively-charged region surrounding a leucine-phenylalanine-proline (LFP) motif, located between the heptad repeat and the first TPR, associates with the TPR-binding groove in competition with WE/D-containing peptides. Inactive KLC thus exists in a folded conformation with its C-terminal tail located such that it could stabilize this folded conformation. Modulatory proteins could either stabilize this conformation or induce a conformational shift facilitating dissociation of the LFP-containing peptide from the WE/D binding groove.
On its own A36 association with KLC is inefficient and may require F12/E2 to provide access to the WE/D binding groove. This model (shown in Figure 8) 58 The reason for targeting a specific KLC iso/spliceform subset could therefore be to utilize efficiently those motor complexes that are available, or to use motors that will take virions to particular locations on the cell surface.
The role of kinesin-1 in virion egress has been studied most extensively in members of the Baculoviridae, Herpesviridae and Poxviridae. 59   HR D in Figure 5.
A plasmid expressing N-terminally V5 (MGKPIPNPLLGLDST)tagged codon optimized VACV E2, pcDNA3-V5E2 (pV5E2) has been described. 40   All site directed mutagenesis was carried out using the Quik-Change II site directed mutagenesis kit (Agilent) and all alleles were sequenced to confirm the presence of the mutations.

| Viruses
All infections were carried out using the Western Reserve (WR) strain of VACV or mutants generated from this strain. Viruses vΔF12, lacking expression of F12, 13