We used five different capsid types and mammalian brain cytosol to analyze the interaction of HSV1 capsids with host MT motors. Starting with nuclei of infected cells or extracellular virions, and varying the salt concentration during virion lysis, we generated capsids with different protein composition and different surface features without impairing their overall architecture. Using immunoblot and quantitative immunoelectron microscopy, we show that dynein, dynactin, kinesin-1 and kinesin-2 bound specifically to viral capsids, but not to nuclear capsids (summarized in & ). Presumably, entire MAP complexes, and not just individual subunits, associated with the viral capsids, since we could detect several subunits of each MAP. In contrast, GST or other cytosolic host proteins such as tau, amphiphysin or LC3 were not recruited.
Scale drawing of a HSV1 viral capsids exposing inner tegument proteins and interacting with MAPs.
colocalizes with inbound cytosolic HSV1 capsids 
. Furthermore, outbound capsids might use dynein for transport to the trans-Golgi network, where HSV1 capsids acquire their final envelope 
, [39; Döhner, Sodeik & Bauerfeind, unpublished observations]
. Our experiments with purified motors showed that dynein can bind directly - independent of other host factors - to tegumented capsids. Similarly dynactin
also bound directly and independently of other host factors to viral capsids. This is different from many host cargos that utilize dynactin as an adaptor to recruit dynein 
. However, destroying the dynactin complex by an artificial excess of its subunit dynamitin reduces nuclear targeting of HSV1 capsids in vivo
as well as the number of motile capsids and the transport velocity in vitro 
. Thus, dynactin may rather enhance dynein processivity than function as an adaptor to recruit dynein onto capsids, as has also been described for some host cargos 
Dynactin also enhances the processivity of kinesin-2
, it is a cofactor for both, dynein and kinesin-2 mediated transport of Xenopus
melanosomes, and it interacts directly with kinesin-2 
. KIF3A, one of the heavy chains of kinesin-2, interacts with pORF45, a tegument protein of the gammaherpesvirus KSHV that has no homolog in alphaherpesviruses. Inhibition of kinesin-2 activity by siRNA gene silencing reduces viral yields of KSHV in lymphoma cells, but not of HSV1 in epithelial cells 
. Thus, so far, there is no evidence of a role for kinesin-2 in the life cycle of alphaherpesviruses. Nevertheless, tegumented, but not nuclear HSV1 capsids specifically recruited kinesin-2, either directly via a tegument protein or via its interaction with dynactin or another host factor (). Kinesin-2 may either facilitate intracellular transport in epithelial cells in a non-essential manner, or catalyze axonal transport of outbound particles in neurons.
also bound directly in the absence of other host factors to 0.5 M KCl capsids. This differs from many membranous host cargos that recruit kinesin-1 via adaptors such as huntingtin-associated protein 1, sunday driver or amyloid precursor protein 
. However, in contrast to tubulin and several proteins of the actin cytoskeleton, such adaptors or other MAP subunits have not been detected in purified herpesvirus particles 
. Kinesin-1 may catalyze HSV1 capsid transport from the MTOC to the nucleus, it colocalizes with outbound HSV1 particles in axons, and it may mediate axonal transport of progeny cytosolic capsids 
. Inhibiting kinesin-1 but not kinesin-2 by siRNA gene silencing or overexpression of dominant-negative protein domains reduced nuclear targeting of inbound HSV1 capsids in epithelial cells (Janus & Sodeik, in preparation).
The number of MT motors
is unknown for viral cargos, but on host cargos ranges from 2 to 10 or even more [reviewed in 7]
. Just one dynein or kinesin-1 motor is sufficient to transport synthetic beads over long distances in vitro
, whereas at high dynein-dynactin concentrations such beads stall at MT intersections 
. Cargo driven by too many motors seems more likely to get trapped in a fruitless tug-of-war between motors interacting with different MTs. Each reaction contained viral capsids derived from 5×108
PFU/60 µL with about 10 to 60 DNAse protected genomes/PFU 
. Taking this number of genomes as the lowest estimate, the viral capsids were within a concentration of 0.1 to 0.8 nM. Based on immunoblots and comparing with protein concentrations of the MAP preparations, we estimate that our cytosolic extracts contain about 30 nM dynein, 40 nM dynactin, and 190 nM kinesin-1. Thus, there were about 35 to 300 molecules dynein, 50 to 400 molecules dynactin, and 240 to 1900 molecules of kinesin-1 present per capsid. Compared to the cytosol loading control, about 1 to 20% of the total MAPs were co-sedimented with viral capsids treated with 0.5 M KCl. These rough estimations suggest that in our assays, a capsid recruited 0.3 to 60 copies of dynein, 0.5 to 80 copies of dynactin, and 2.5 to 380 copies of kinesin-1.
On thawed cryosections of cells, protein-A gold labels antibodies directed against dynein on about 13% of inbound cytosolic HSV1 capsids 
. The labeling efficiency for this technique is estimated to detect 5 to 10% of the antigens present 
. Thus, it seems reasonable to assume that, in cells, all incoming capsids are associated with dynein. Here we labeled for host factors bound to capsids in vitro
which should be even better accessible to antibodies, although the antibodies and protein-A gold may still preferentially label the apical capsid surfaces over those facing the substrate. Our immunoelectron microscopy demonstrated that at least 20% of the capsids accommodated more than one copy of dynein, and about 10% more than one dynactin or kinesin-1 complex.
Since dynein and dynactin, but not kinesin-1, bound to 1 M KCl capsids (, Supplementary Table S1
), and given that purified dynein and dynactin bound independently of each other, these 3 MAPs seem to associate with different surface features or domains on 0.5 M KCl capsids. As nuclear capsids did not bind MAPs, the tegument and not the capsid proteins must contain the viral receptors for microtubule motors
. Capsids derived from HSV1 knock-out strains showed that VP26, pUS11 or VP11/12 do not contribute to the generation of the viral motor receptors that we analyzed here. Furthermore, PrV does not encode a homolog of US11, and HSV1 mutants lacking VP26 or pUS11, as well as PrV lacking VP26, pUL14, VP11/12 or VP22 are not impaired in nuclear targeting of capsids 
, [69; Janus & Sodeik, in preparation]
. Thus, these viral proteins are less likely to contribute to the generation of viral motor binding domains.
Interestingly, capsids derived from a HSV1 mutant lacking VP11/12 bound even more MAPs than wild-type. As dynein, dynactin, kinesin-1 or kinesin-2 are not present in complete virions 
, the acquisition of outer tegument proteins during assembly may gradually compete for MAP binding. Since outer tegument proteins associate with viral envelope proteins 
, MAPs may even need to dissociate prior to secondary envelopment. A reverse reaction seems to occur during cell entry. Upon viral fusion, the bulk of tegument proteins detaches from incoming HSV1 and PrV capsids and remains bound to the viral envelope, thereby exposing the underlying capsid-associated inner tegument proteins 
. We suppose that treating virions with TX-100 and 0.5 M KCl results in a similar separation of the outer from the inner tegument layer, and thus enables access of cytosolic host factors to capsid-associated viral motor receptors. This notion is supported by the low binding of MAPs to capsids treated with 0.1 M KCl that still contain high amounts of the tegument proteins vhs, pUL11, ICP4, ICP34.5, VP11/12, VP13/14, VP16, and VP22. We therefore classified proteins that were abundant on 0.1 M capsids, but reduced by treatment with 0.5 or 1.0 M KCl as outer tegument proteins (; Supplementary Table S1
). This outer tegument rather prevented than mediated a recruitment of MT motors to herpesvirus capsids.
The 0.1 and 0.5 M KCl treated capsids contained other proteins in similar amounts that we consider inner tegument components: pUS3, pUL36, pUL37, ICP0, pUL14, pUL16, and pUL21 (, Supplementary Table S1
). While pUL21 is not required for retrograde spread of PrV, the tegument is not properly assembled in a PrV mutant lacking UL21 
. Thus, binding assays using capsids derived from such mutants alone would not suffice to demonstrate that pUL21 acts as a viral receptor for motors. pUL36, pUL37, or pUL21 may interact with kinesin-1 or at least contribute to the formation of a viral kinesin-1 receptor, since these proteins were reduced or at least some of their epitopes were denatured by treating viral capsids with 1 M KCl (, Supplementary Table S1
Considering all HSV1 structural proteins, pUL36 and pUL37
are likely candidates to either directly enlist motors to capsids, or to recruit other tegument proteins that function as viral motor receptors. pUL37 interacts directly with the capsid protein VP26, and with pUL36 that is linked via pUL25 to capsids 
. pUS3, pUL36, and pUL37GFP remain associated with incoming capsids during nuclear targeting; and HSV1 and PrV mutants lacking pUL36 or pUL37 are impaired in intracellular transport 
, [83; Schipke & Sodeik, in preparation]
. We show here that pUL36 and pUL37GFP were exposed on the surface of 0.5 M KCl treated capsids, and that epitopes of pUL36 and pUL37 were altered by increasing KCl concentration from 0.5 to 1 M which also inactivated the receptor binding site for kinesin-1 (, Supplementary Table S1
Unfortunately, capsids of HSV1 or PrV mutants lacking pUL36 or pUL37 cannot be isolated from extracellular virions, since cytosolic capsids are not enveloped, and no virions are assembled 
. Therefore, binding assays with capsids derived from such mutants require new purification protocols to isolate sufficient amounts of cytosolic progeny capsids not contaminated by the more abundant nuclear capsids 
. Considering the multivalent interactions among several tegument proteins 
, the lack of pUL36 or pUL37 may result in an altered tegument organization, as has already been shown for mutants lacking pUL21 
. Future binding studies with such mutant capsids will therefore require a thorough investigation of their molecular tegument architecture.
The capsid transport direction
to the cell center or periphery during cell entry and egress must be regulated. Our data suggest that dynein and kinesin-1 bind to distinct surface features on 0.5 M KCl capsids, since treatment with 1 M KCl specifically inactivated kinesin-1 but not dynein binding. A similar modification of the tegument surface may also occur in vivo
and change the affinity of capsids for motors, e.g. during cell entry when pUL36 is cleaved, or tegument proteins are phosphorylated 
. Thus, an “exclusive” presence of either minus- or plus-end directed motors on viral capsids of different protein composition may regulate their transport direction 
, while such a mechanism seems to be less used for host cargos 
. Time-lapse microscopy shows that HSV1 and PrV particles quickly change transport directions in vivo
suggesting that motors of opposing directionality are present simultaneously on viral particles. Nuclear targeting of inbound and egress of outbound PrV particles can occur simultaneously within the same cell, suggesting that there is no overall inhibition or stimulation of host factors controlling transport direction during herpesvirus infection 
. However, in many live cell microscopy experiments, transport of cytosolic capsids cannot be distinguished from transport of virions within host membranes 
, [86; Janus & Sodeik, in preparation]
Our biochemical single-particle analysis shows that at least 10% of all capsids had bound the both motors dynein and kinesin-1 simultaneously. Thus, it seems likely that also in cells individual cytosolic capsids can recruit the entire transport machinery of dynein, dynactin, and kinesin-1 (). Therefore, the transport direction of HSV1 capsids does not seem to be regulated by an exclusive presence of either plus- or minus-end directed motors. The bound motors rather either engage in a tug-of-war to determine the direction of net transport, or the activities of the bound motors are coordinately regulated. For example, an alternating activation of either plus- or minus-end directed motors e.g. by host or viral kinases or binding of cellular inhibitors could mediate fast changes in transport direction.
Future studies with such biochemical assays will provide further insights into the regulation of essential virus-host interactions, for example by adding, depleting, or inhibiting specific host factors such as kinases. Furthermore, the functions of structural viral proteins can be elucidated even if their mutation prevents virus formation or cell culture models are lacking. Recombinant proteins, protein fragments, or peptides derived from viral genes can be tested for their ability to compete with particle-associated viral proteins for the recruitment of host factors. Such assays will identify viral structures that are required for motor binding and capsid nuclear targeting, and that therefore should be maintained in viral gene therapy vectors. These structural features or domains may even be added to artificial nano carriers or other vectors to be exploited for human therapeutic gene and cell therapy.