TI - Results . AB - The characterization of lipid rafts has largely been based on their resistance to detergent solubilization at 4degC [3] . This approach continues to be a useful tool to assess the affinity a protein has for lipid rafts when formed at physiologic temperatures [38] , [39] . Analyses performed on changes in DRM partitioning ( due to a physiological stimulus ) have provided the foundation for extensive work on the role of lipid rafts in signal transduction [40] -[42] , membrane trafficking [43] , [44] and pathogenesis [45] . Accordingly , we investigated the affinity of Us9 for DRMs , as well as several viral glycoproteins whose axonal localization is dependent on Us9 [16] . It is difficult to culture a sufficient number of primary rat neurons to perform large biochemical analyses . Therefore , we used PC12 cells , a widely used rat pheochromocytoma cell line that responds to nerve growth factor ( NGF ) and acquires many of the characteristics of sympathetic neurons [46] . Differentiated PC12 undergo polarized protein sorting , and cell bodies stain for nonphosphorylaTED neurofilament H ( a somatodendritic marker ) while axons stain exclusively for PHOSphorylated neurofilament H ( an axonal marker ) [32] . This is consistent with mature , sympathetic SCG neurons [47] . It has been reported that PC12 cell are susceptible and permissive to PRV infection [48] , and that a PRV GFP-VP22 fusion protein moves inside neurites with fast axonal kinetics [49] . However , it was unclear whether the Us9-null phenotype in SCG neurons , i.e.a complete block to axonal sorting of viral structural proteins [15] , [16] , could be recapitulated in this neuron-like cell line ( a critical experiment to ensure that PC12 cells could be used to study Us9 biology ) . We recently reported that in the absence of Us9 , GFP-tagged capsids were unable to sort into axons of live SCG neurons [15] . Therefore , we utilized a similar live cell imaging approach to examine the axonal sorting of GFP-tagged capsids in differentiated PC12 cells . Cells were infected with PRV GS443 , a recombinant PRV strain that expresses GFP fused to VP26 , a capsid protein [50] . After 12 h , capsid puncta were readily observed trafficking in the anterograde direction within neurites of PC12 cells ( n = 20 ) ( Figure 1A , Video S1 ) . Importantly , when differentiated PC12 cells were infected with PRV 368 , a GFP-tagged capsid mutant deleted for Us9 , no green puncta were observed moving in the anterograde direction ( Figure 1B and 1C , Video S2 ) . These findings were consistent with our Us9 studies in dissociated SCG neurons [15] . Interestingly , we also observed the retrograde trafficking of capsid puncta from cells infected with PRV 368 to uninfected , neighboring cells ( in the absence of any anterograde sorting of virus particles in the same field of view ) ( Figure 1D-1F , Videos S3 and S4 ) . This had been described previously in transneuronal spread studies on Us9 mutants in the rat visual system [11] , [26] , but had not been observed in tissue culture cells . It is noteworthy that we did not visualize "random" egress of GFP-tagged capsids from infected cell bodies . Capsids either sorted into axons ( in the presence of Us9 ) , or to sites of synaptic contact with other axons ( transneuronal , retrograde transport ) . The import of this observation is unclear at present , but may suggest that alpha herpesviruses undergo directed egress from neuronal cell bodies . Overall , our findings suggest that differentiated PC12 cells recapitulate the Us9 sorting phenotypes previously observed in primary sympathetic neurons , and are an efficacious cell line to study Us9 biology ( specifically that of Us9 and lipid rafts ) . Therefore , we compared the raft profiles of viral membrane proteins after infection of non-polarized and polarized PC12 cells . Undifferentiated cells ( ~107 ) were infected with wild-type Becker for 12 hours , solubilized with 1% TX-100 , and subjected to a well-described "raft flotation" assay [33] -[35] . Membrane proteins that were solubilized by TX-100 remained at the bottom of the Opti-Prep gradient ( 40% ) , whereas proteins in DRMs floated to the 5%-30% interface . We used the prototypic raft and non-raft markers , GM1 ganglioside and transferrin receptor ( TfR ) , as positive and negative controls [51] . Us9 was highly enriched in the raft fraction as compared to the soluble population ( Figure 2 ) . To test whether the flotation of Us9 was cholesterol dependent , we treated cells with methyl-cyclodextrin ( MCD ) prior to solubilization with detergent . MCD depletes cholesterol from cellular membranes , and therefore disrupts the structure of lipid raft microdomains [52] , [53] . A 45-minute exposure of 20 mM MCD to Becker-infected cells dramatically decreased the amount of Us9 floating with the raft fraction . We found the viral glycoprotein gB to have a strong affinity for DRMs while gE and gC where not enriched in either the raft or soluble fractions . These findings were consistent with similar experiments performed in non-polarized swine kidney ( SK ) cells [18] . Surprisingly , PRV gH was completely solubilized by TX-100 treatment , as was transferrin receptor ( the negative control ) . Overall , Us9 and gB have a strong affinity for DRMs in undifferentiated PC12 cells . Both gE and gC were present in the raft and soluble fractions in a ~1[? lt @@@@@ gt ]1 ratio , whereas gH was completely soluble . To determine the effect of cell polarization on DRM partitioning , we cultured PC12 cells in low serum conditions in the presence of nerve growth factor for 12 days . This treatment allows for extensive neurite outgrowth from the cell bodies , as well as separation of somatodendritic and axonal marker proteins [32] . We infected cells with PRV Becker for 12 hours , and subjected lysates to raft flotation analysis as performed previously ( Figure 3 ) . Us9 and gB were again strongly associated with the raft fraction of the gradient , and floated with GM1 . Both gE and gC were highly enriched in the DRM fraction of PC12 cells upon differentiation with NGF . This finding is consistent with the notion that polarization of neurons strongly impacts the affinity and targeting of certain membrane proteins to lipid rafts/DRMs [4] . gH was completely solubilized by TX-100 even in differentiated PC12 cells , and remained in the soluble fraction with TfR . These data suggest that Us9 and gB have a high affinity for DRMs in undifferentiated and differentiated PC12 cells , whereas gC and gE increase their association with DRMs as cells polarize and mature . The efficient targeting of viral structural components to the axon of infected cells is dependent on both the Us9 and gE gene products [15] , [16] , [54] . Deletion of either gene results in the reduction of viral capsids and enveloped proteins in the axon , and subsequent reduction of anterograde spread of infection in vitro and in vivo[11] , [17] , [55] . Upon discovering that both Us9 and gE were present in the DRMs of infected PC12 cells , we tested whether deleting the gE/gI complex affected the ability of Us9 to target to rafts , thereby impacting its ability to function properly in anterograde transport . PC12 cells were infected with PRV 99 , a mutant deleted for the gE and gI genes , and subjected to raft flotation analysis . Deletion of gE/gI had no affect on the ability of Us9 to associate with DRMs (Figure 4A) , nor did the absence of gB ( data not shown ) . These data again support the notion that Us9 has an intrinsic affinity for DRMs and is not influenced by cell polarity or the presence of two major viral DRM components . It is noteworthy that we found no aberrant targeting of gB , gC or gE to DRMs in a Us9-null mutant ( data not shown ) . It is well documented that certain proteins become raft associated upon their PHOSphorylation during signal transduction [56] -[58] . Us9 contains a conserved acid cluster ( AC ) region with two serines that are PHOSphorylated [26] , as well as a di-tyrosine motif critical for anterograde , transynaptic spread [26] . To examine whether inhibiting Us9 PHOSphorylation precluded Us9 association with DRMs , we infected PC12 cells with PRV 162 , a mutant that expresses an altered Us9 protein lacking the acidic cluster region . DRMs were prepared from infected cells as performed previously ( Figure 4B ) . Us9 was still enriched in the DRM fraction of the gradient despite the absence of Us9 PHOSphorylation ( note the narrowness of the Us9 band compared to wild-type Us9 in panel A ) . Taken together , these data suggest that Us9 is highly enriched in DRMs , and its affinity for this lipid microdomain is not dependent on its PHOSphorylation state . Both the influenza virus neuraminidase ( NA ) and haemagglutinin ( HA ) proteins ( both type II membrane proteins ) are highly enriched in DRMs/lipid rafts , and this association is critical for the apical sorting of these proteins in polarized MDCK cells [59] -[61] . Importantly , the transmembrane domain ( TMD ) of these proteins provided the determinants for apical sorting and raft association [59] , [62] -[64] . This was demonstrated by swapping the TMD of transferrin receptor ( a type II , non-raft associated membrane protein ) for the TMD of neuraminidase [63] . Transferrin receptor was normally sorted to the basolateral membrane in polarized MDCK cells , and was efficiently solubilized by 1% TX-100 . By contrast , transferrin receptor with the NA TMD was targeted to the apical cell surface and was largely insoluble to treatment with TX-100 [59] . In a reciprocal experiment , the neuraminidase TMD was replaced with the transferrin receptor TMD [62] . This chimeric protein was greatly reduced in lipid raft association , and a virus expressing this protein showed a defect of particle release from the apical cell surface . We employed a similar approach with PRV Us9 to test whether its transmembrane domain provided the determinants for raft sorting . Both Us9 and transferrin receptor are type II membrane proteins , and have 26 amino acids within their TMD . We constructed a PRV mutant that expressed a chimeric protein with the wild-type Us9 cytoplasmic domain , a transferrin receptor TMD and wild-type 3 amino acid ectodomain (Figure 5A) . This mutant , known as PRV 322 , replicates with wild-type kinetics in porcine kidney ( PK15 ) cells . Furthermore , the Us9-TfR protein is abundantly expressed in infected cell lysates and migrates more slowly by SDS-PAGE than wild-type Us9 (Figure 5C) . Expression of the upstream and downstream genes , gE and Us2 respectively , were indistinguishable from Becker and suggested that recombination of Us9-TfR into the viral genome did not have polar effects on neighboring genes . Us9-TfR is efficiently incorporated into virions ( Figure 5D ) , along with gE and Us2 , which is a component of primary but not mature virus particles [65] . The localization and intracellular trafficking pattern of Us9 has been studied extensively in porcine kidney ( PK15 ) cells [20] , [26] , [28] . To assess whether Us9-TfR remained a functional Us9 derivative ( ie did not have an egregious trafficking defect ) , we examined the steady-state localization of Us9-TfR in transfected and infected PK15 cells . We hypothesized that both Us9 and Us9-TfR would have similar localization patterns as both contain the Us9 acidic domain , the region necessary for localization to a perinuclear cellular compartment [20] . Furthermore , sorting signals for the TfR protein reside in the cytoplasmic tail , not within the transmembrane domain [66] . Thus , the TfR TMD should be functionally inert in the context of Us9 trafficking . The localization of Us9 fused to GFP mimics the localization of wild-type Us9 inside infected cells [20] , [28] . We fused Us9-TfR to GFP to visualize its steady-state level in cells in the absence of infection . Both Us9-GFP and Us9-TfR-GFP were predominantly located to a perinuclear region in the cytoplasm of PK15 cells (Figure 6A) . Confocal images of cells infected with Becker and PRV 322 , fixed and stained with Us9 antiserum , also showed co-localization of Us9 and Us9-TfR to a perinuclear compartment , though the Us9-TfR signal was slightly more diffuse compared to the Us9 signal (Figure 6B) . Overall , the trafficking of Us9-TfR inside transfected and infected cells was very similar to wild-type Us9 . Next we tested whether the replacement of the wild-type Us9 TMD with that of transferrin receptor affected the affinity of Us9 for DRMs . We infected undifferentiated and differentiated PC12 cells for 12 hours , solubilized cells with 1% TX-100 , and performed flotation analysis as done previously . Instead of Us9 being heavily enriched in the raft fraction as observed in Becker infected cells ( see Figures 2 and 3 ) , Us9-TfR was predominantly found in the soluble fraction , especially in differentiated PC12 cells ( Figure 7A and 7B ) . To assess whether this impacted Us9 function in primary neurons , we took advantage of a trichamber neuronal culturing system [17] , [37] . Dissociated SCG neurons are plated in the soma (S) chamber and allowed to mature for two weeks ( Figure 7C ) . During this period , axons are directed between a series of grooves across the methocellulose (M) chamber to the neurite (N) chamber . A monolayer of indicator PK15 cells are then plated on top of the neurites in the N chamber . Cell bodies in the S chamber are infected , virus particles sort into axons in a Us9 -dependent manner , and subsequently infect the PK15 cells that amplify the infection . The initial infection is confined to the S chamber via silicone vacuum grease and a methocellulose barrier . Therefore , infection spreads to the N chamber solely through axons that emanate from neuronal cell bodies and extend to PK15 cells [17] . We compared the anterograde transport and spread capabilities of PRV Becker ( wild-type ) , PRV 160 (Us9-null) , PRV 322 (Us9-TfR) , and a co-infection of Becker and PRV 322 ( Figure 7C , lower panel ) . Though all of the infections produced a comparable number of infectious virus in the S chamber , spread to second order PK15 cells in the N chamber was dramatically different . PRV Becker spread efficiently from neurons to PK15 cells , producing a median titer of 1.2x107 PFU in the N chamber after 24 hours post-infection ( Figure 7C ) . By contrast , the Us9-null mutant (PRV 160) did not spread to PK15 cells and no detectable infectious virus was produced in most dishes . However , in one dish , we detected a low number of infectious particles ( 15x103 ) . We interpret a low yield of amplified virus as a single neuron-to cell spread event ( the burst size of an infected PK15 cell is roughly 1000 PFU ) . Nevertheless , the neuron-to cell spread capability of PRV 160 is extremely low compared to wild-type PRV Becker . PRV 322 (Us9-TfR) was completely defective in anterograde spread and was indistinguishable from the Us9-null mutant ( no infectious virus detected in the N-compartment ) . This phenotype strongly correlated with the inability of Us9-TfR to target to lipid rafts/DRMs . When neurons were co-infected with both Becker and PRV 322 , titers were virtually identical to those seen with Becker alone , indicating that PRV 322 does not have a trans-dominant effect on anterograde spread of infection . To assess whether the anterograde , neuron-to cell spread defect for PRV 322 was at the level of axonal sorting of viral structural proteins ( as previously shown for other Us9 mutants [15] , [16] ) , we imaged infected neurons in the trichamber system using a PRV-specific antibody ( made against acetone-fixed virus particles ) that recognizes both virus glycoproteins and virus capsid proteins [21] , [22] . PRV antigen was readily detected in the cell bodies of neurons in the S compartment infected with Becker , PRV 160 and PRV 322 ( Figure 8 , first column ) . Viral glycoprotein and capsid proteins were also abundant in the axons of Becker infected neurons within the N compartment ( Figure 8 , second column ) . By contrast , no viral structural proteins were observed by immunofluorescence in the axons of a Us9-null mutant (PRV 160) or the Us9-TfR strain (PRV 322) , though an extensive network of axons was observed within the field of views by transmitted brightfield illumination ( Figure 8 , transmitted ) . These data suggest that the neuron-to cell spread defect observed for PRV 322 ( Figure 7 ) is the result of its inability to sort structural proteins into the axon of infected neurons . Overall , these data are consistent with work done on the raft association of the influenza virus neuraminidase protein : substitution of a TMD domain that has a high affinity for lipid rafts , for one with a low affinity , dramatically alters DRM targeting and subsequent protein function .