US3 Kinase-Mediated Phosphorylation of Tegument Protein VP8 Plays a Critical Role in the Cellular Localization of VP8 and Its Effect on the Lipid Metabolism of Bovine Herpesvirus 1-Infected Cells

Nuclear localization signals (NLSs) and nuclear export signals (NESs) are important elements directing VP8 to the desired locations in the BoHV-1-infected cell. In this study, a critical regulator that switches the nuclear and cytoplasmic localization of VP8 in BoHV-1-infected cells was identified. BoHV-1 used viral kinase US3 to regulate the cellular localization of VP8. Early during BoHV-1 infection VP8 was localized in the nucleus, where it performs various functions; once US3 was expressed, phosphorylated VP8 was cytoplasmic and ultimately accumulated in the cis-Golgi apparatus, presumably to be incorporated into virions. The Golgi localization of VP8 was only observed in virus-infected cells and not in US3-cotransfected cells, suggesting that this is mediated by other viral factors. Interestingly, VP8 was shown to cause increased cholesterol levels, which is a novel function for VP8 and a potential strategy to supply lipid for viral replication.

ABSTRACT

Bovine herpesvirus 1 (BoHV-1) infects bovine species, causing respiratory infections, genital disorders and abortions. VP8 is the most abundant tegument protein of BoHV-1 and is critical for virus replication in cattle. In this study, the cellular transport of VP8 in BoHV-1-infected cells and its ability to alter the cellular lipid metabolism were investigated. A viral kinase, US3, was found to be involved in regulating these processes. In the early stages of infection VP8 was localized in the nucleus. Subsequently, presumably after completion of its role in the nucleus, VP8 was translocated to the cytoplasm. When US3 was deleted or the essential US3 phosphorylation site of VP8 was mutated in BoHV-1, the majority of VP8 was localized in the nuclei of infected cells. This suggests that phosphorylation by US3 may be critical for cytoplasmic localization of VP8. Eventually, the cytoplasmic VP8 was accumulated in the cis-Golgi apparatus but not in the trans-Golgi network, implying that VP8 was not involved in virion transport toward and budding from the cell membrane. VP8 caused lipid droplet (LD) formation in the nuclei of transfected cells and increased cellular cholesterol levels. Lipid droplets were not found in the nuclei of BoHV-1-infected cells when VP8 was cytoplasmic in the presence of US3. However, when US3 was deleted or phosphorylation residues in VP8 were mutated, nuclear VP8 and LDs appeared in BoHV-1-infected cells. The total cholesterol level was increased in BoHV-1-infected cells but not in ΔU_L47-BoHV-1-infected cells, further supporting a role for VP8 in altering the cellular lipid metabolism during infection.

IMPORTANCE Nuclear localization signals (NLSs) and nuclear export signals (NESs) are important elements directing VP8 to the desired locations in the BoHV-1-infected cell. In this study, a critical regulator that switches the nuclear and cytoplasmic localization of VP8 in BoHV-1-infected cells was identified. BoHV-1 used viral kinase US3 to regulate the cellular localization of VP8. Early during BoHV-1 infection VP8 was localized in the nucleus, where it performs various functions; once US3 was expressed, phosphorylated VP8 was cytoplasmic and ultimately accumulated in the cis-Golgi apparatus, presumably to be incorporated into virions. The Golgi localization of VP8 was only observed in virus-infected cells and not in US3-cotransfected cells, suggesting that this is mediated by other viral factors. Interestingly, VP8 was shown to cause increased cholesterol levels, which is a novel function for VP8 and a potential strategy to supply lipid for viral replication.

INTRODUCTION

Bovine herpesvirus 1 (BoHV-1) is a member of the Alphaherpesvirinae and predominantly infects cattle. It causes infectious bovine rhinotracheitis, pustular vulvovaginitis, and balanoposthitis (1). This virus may also cause abortions in host animals during pregnancy (2). The double-stranded DNA genome of BoHV-1 is 135 kbp and is enclosed in a capsid shell, which is about 125 nm in diameter (3). Outside the capsid is a tegument protein layer surrounded by a lipid envelope and glycoproteins (4). VP8 is the major component of the tegument and essential for BoHV-1 to infect host animals (4, 5). It is a late protein expressed by the U_L47 gene, which is conserved in the Alphaherpesvirinae (6). For example, in human herpesvirus 1 (HHV-1) the U_L47 gene expresses a nonessential tegument protein, named VP13/14 (7).

VP8 is phosphorylated by the viral unique short protein 3 (US3) and the cellular casein kinase 2 (CK2) in BoHV-1-infected cells (8, 9). Virion VP8 is dephosphorylated, indicating that the major role of phosphorylation might be regulating cellular functions of VP8 (8). US3 phosphorylates VP8 at residues S¹⁶ and S³² (8). Phosphorylation at S¹⁶ is essential for the subsequent phosphorylation at S³² (8). CK2 has multiple targets on VP8 with different preferences. Seven residues (T⁶⁵, S⁶⁶, S⁷⁹, S⁸⁰, S⁸², S⁸⁸, and T¹⁰⁷) in the N terminus of VP8 are critical for phosphorylation through CK2, T¹⁰⁷ being most frequently phosphorylated (8).

The cellular localization of VP8 changes with the progression of BoHV-1 infection (5). Early during infection, VP8 is mostly in the nucleus. The nuclear localization of VP8 is mediated by arginine-rich nuclear localization signal 1 (NLS1; P¹¹RPRR¹⁵) and NLS2 (R⁴⁸PRVRRPRP⁵⁴) (10, 11). Subsequently, VP8 is exported into the cytoplasm and accumulates in the Golgi apparatus at later stages of infection (12). At least two nuclear export signals (NESs) have been described for VP8. One of them is a chromosomal maintenance 1 (CRM1)-dependent NES, and the other one is a CRM1-independent NES (13). It has been suggested that they are not the only NESs in VP8 because mutating both NESs does not completely block VP8 translocation from one nucleus to another within the same cell generated by interspecies heterokaryons (10, 13). The NLSs and NESs of VP8 might be regulated as a viral strategy to precisely navigate VP8 to different subcellular locations at different stages of the BoHV-1 life cycle.

Phosphorylation-regulated localization of proteins has been reported for cellular and viral proteins. For example, phosphorylation and dephosphorylation control the subcellular transport of eukaryotic translation initiation factor 6 (eIF6), a protein that is essential for the separation of the 60S subunit from the 40S subunit (14). When eIF6 is phosphorylated through casein kinase 1 (CK1), it is translocated from the nucleus to the cytoplasm along with the 60S subunit (14). The cytoplasmic eIF6 is then dephosphorylated through calcineurin (14) and subsequently recycled to the nucleus (15). Phosphorylation also controls the subcellular localization of VP13/14 in HHV-1 (6). The nuclear localization of VP13/14 is mediated through an NLS and is regulated by US3 of HHV-1. US3-phosphorylated VP13/14 localizes in the nucleoplasm and in the nuclear membrane in HHV-1-infected cells. However, nonphosphorylated VP13/14 is predominantly in the nuclear membrane. This translocation of VP13/14 is correlated to stromal keratitis caused by HHV-1 in mice (16). The phosphoprotein VP8 of BoHV-1 has been described as a nuclear-cytoplasmic shuttling protein, leading to a hypothesis that the cellular localization of VP8 might be regulated by US3- and/or CK2-mediated phosphorylation.

RESULTS

Nuclear VP8 is transported to the cytoplasm during the late phase of BoHV-1 infection.

While in the nucleus early during infection, VP8 was found to accumulate in the Golgi apparatus in BoHV-1-infected cells late during infection (12). This raised the question of whether the previously synthesized nuclear VP8 or the newly synthesized cytoplasmic VP8 was accumulated in the Golgi. To determine the source of Golgi-localized VP8, wild-type (WT) VP8 with two different tags (FLAG-VP8 and yellow fluorescent protein [YFP]-VP8) was expressed in embryonic bovine tracheal (EBTr) cells, a bovine cell line that is susceptible to BoHV-1 infection and to transient transfection. FLAG-VP8 was expressed by transient transfection. After 24 h, the transfected cells were mock infected or infected with BoHV-1-YVP8 to express YFP-VP8. Transiently expressed FLAG-VP8 localized in the nuclei of mock-infected cells, which did not express YFP-VP8, from 28 h posttransfection (hpt) to 43 hpt (Fig. 1). In comparison, when transfected cells were infected with BoHV-1-YVP8, YFP-VP8 appeared in the nucleus and colocalized with FLAG-VP8 at 4 h postinfection (hpi) (Fig. 1). At 11 hpi YFP-VP8 appeared in the nuclear periphery of the cytoplasm and nuclear YFP-VP8 was reduced. At 19 hpi most of the YFP-VP8 was accumulated in the perinuclear region of the cytoplasm, while nuclear YFP-VP8 was greatly reduced. Within the time frame of infection, transfected FLAG-VP8 was translocated from the nucleus to the cytoplasm in a fashion identical to that of YFP-VP8, providing evidence that nuclear VP8 is transported to the Golgi during infection.

FIG 1.

Translocation of VP8 from the nucleus to the cytoplasm in BoHV-1-YVP8-infected cells. EBTr cells transfected with pFLAG-VP8 were cultured for 24 h. The FLAG-VP8-expressing cells were then mock infected or infected with BoHV-1-YVP8 at an MOI of 5. Cells were fixed at the indicated time points for immunofluorescent staining. FLAG-VP8 was detected with monoclonal anti-FLAG antibody followed by Alexa Fluor 633-conjugated goat anti-mouse IgG. YFP-VP8 was tracked through the YFP label. The cell nuclei were identified with DAPI.

US3 is critical for the cytoplasmic translocation of VP8 during a late stage of BoHV-1 infection.

To examine the impact of US3 on the cellular localization of VP8, Madin-Darby bovine kidney (MDBK) cells, a bovine cell line susceptible to BoHV-1 infection, were infected with wild-type BoHV-1 (Cooper and 108), ΔUS3-BoHV-1, or RUS3-BoHV-1. The expression of US3 and the localization of VP8 were examined at 4 and 8 hpi. US3 was localized in the nuclei of wild-type BoHV-1 and RUS3-BoHV-1 infected cells, while ΔUS3-BoHV-1-infected cells did not express US3. At 4 hpi VP8 appeared in the nuclei of cells infected with wild-type viruses, ΔUS3-BoHV-1, and RUS3-BoHV-1 (Fig. 2A). There was no noticeable difference between the four viruses in the nuclear localization of VP8 early during infection. At 8 hpi, VP8 was translocated to the cytoplasm of wild-type and revertant virus-infected cells. The wild-type and revertant viruses showed comparable patterns of VP8 localization, and the expression of US3 was increased in these cells. However, in ΔUS3-BoHV-1-infected cells, which did not express US3, the majority of VP8 was in the nuclei with a punctate appearance (Fig. 2B). VP8 and US3 were not observed in mock-infected cells (Fig. 2C). The protein expression by the four viruses was confirmed by Western blotting (Fig. 2D), showing that VP8 was expressed by all viruses and that US3 was expressed by wild-type and revertant viruses but not by ΔUS3-BoHV-1. US3 of the Cooper strain migrated slightly faster than US3 of the 108 strain. Sequencing results showed that K¹⁰⁷ and K¹⁰⁹ in US3 of the Cooper strain were replaced with E¹⁰⁷ and E¹⁰⁹ in the 108 strain, while the rest of the sequences were identical (Fig. 3), indicating that these mutations may change posttranslational modifications in US3, which subsequently affect the migration rate of US3 in an acrylamide gel as shown in Fig. 2D. However, the kinase activity of US3 is unlikely to be affected because the catalytic loop and the ATP-binding pocket are conserved (Fig. 3) (17). These results imply that US3 is essential for cytoplasmic localization of VP8 at a later stage during BoHV-1 infection.

FIG 2.

Cytoplasmic localization of VP8 at a late stage of BoHV-1 infection requires US3. (A to C) MDBK cells were infected with the indicated viruses at an MOI of 5 or mock-infected. Cells were processed for immunofluorescence at 4 and 8 hpi. VP8 was detected with monoclonal anti-VP8 antibody followed by Alexa Fluor 488-conjugated goat anti-mouse IgG. US3 was identified with polyclonal anti-US3 antibody followed by Alexa Fluor 633-conjugated goat anti-rabbit IgG. The cell nuclei were identified with DAPI. (D) Virus-infected MDBK cell lysates were analyzed by Western blotting. VP8 and US3 were detected with mouse anti-VP8 and rabbit anti-US3 antibodies followed by IRDye 800CW goat anti-mouse IgG and IRDye 680RD goat anti-rabbit IgG, respectively.

FIG 3.

Sequence analysis of US3 of BoHV-1 strains 108 and Cooper. The full-length US3 genes amplified from the viral genomes of strains 108 and Cooper were sequenced. The DNA sequences were then translated into amino acid sequences. The two protein sequences were identical to those corresponding to A0A1P8L092 and Q76PF3 in a protein bank, UniProKB (http://www.uniprot.org/). The partial sequences containing mutated residues, an ATP-binding pocket, and a catalytic loop are shown.

Phosphorylation by US3, but not by CK2, correlates with cytoplasmic accumulation of VP8.

US3 was required for cytoplasmic localization of VP8. To determine the reason for this observation, the translocation of wild-type VP8 was examined in transfected cells expressing either wild-type US3 or a mutated US3 (US3-K282E) (17), which does not phosphorylate VP8 (8). VP8 was mainly localized in the nucleus when US3 was absent or mutated (Fig. 4A, panels 1 and 3), while VP8 was localized in the cytoplasm and the nucleus when wild-type US3 and VP8 were cotransfected (Fig. 4A, panel 2). In addition, wild-type US3 was cotransfected with VP8-S16A, which is not phosphorylated by US3 (8). VP8-S16A localized in the nucleus in the absence or in the presence of wild-type US3 (Fig. 4A, panels 4 and 5). The cytoplasmic VP8 and nuclear VP8 were quantified based on the intensity of green fluorescence pixels within the confocal microscope pictures of the above-described experiments (Fig. 4B). In cells cotransfected with wild-type US3 and wild-type VP8, the cytoplasmic intensity of VP8 was significantly higher than in cells not expressing US3. Cotransfection of US3-K282E did not increase the cytoplasmic VP8, suggesting that the kinase activity of US3 was important for the presence of VP8 in the cytoplasm. Similarly, cotransfection of wild-type US3 did not increase the cytoplasmic signal of VP8-S16A compared to that with the single transfection of VP8-S16A, confirming that the nuclear localization of nonphosphorylated VP8 was not affected by US3. The nuclear and cytoplasmic fractions were extracted and the presence of VP8 was analyzed by Western blotting (Fig. 4C). The amount of VP8 increased in the cytoplasmic fractions only when wild-type US3 and wild-type VP8 were cotransfected into cells and not when VP8-S16A or US3-K282E was used. These results confirm a critical role of phosphorylation by US3 in the localization of VP8 to the cytoplasm.

FIG 4.

US3-mediated phosphorylation promotes the cytoplasmic localization of VP8. (A) EBTr cells were transfected with different (combinations of) plasmids. Cells in panel 1 were transfected with pFLAG-VP8. Cells in panel 2 were cotransfected with pFLAG-VP8 and pUS3-HA. Cells in panel 3 were transfected with pFLAG-VP8-S16A. Cells in panel 4 were cotransfected with pFLAG-VP8-S16A and pUS3-HA. VP8 was identified with monoclonal anti-VP8 antibody followed by Alexa Fluor 488-conjugated goat anti-mouse IgG, and US3 was identified with polyclonal anti-US3 antibody followed by Alexa Fluor 633-conjugated goat anti-rabbit IgG. The cell nuclei were identified with DAPI. (B) Relative quantification of cytoplasmic and nuclear VP8. The software Leica LAS AF Lite was used to analyze confocal pictures. The intensity of VP8 is represented as the fluorescence pixels at a wavelength of 488 nm. The mean PDU are shown in the bar graphs. Error bars represent the SDs. The statistical significance is shown as follows: **, P ≤ 0.01. (C) The nuclear and cytoplasmic fractions were extracted from EBTr cells transfected with the indicated plasmids. VP8 and US3 in the nuclear and cytoplasmic fractions were analyzed by Western blotting. VP8 and US3 were detected with mouse anti-VP8 and rabbit anti-US3 antibodies followed by IRDye 800CW goat anti-mouse IgG and IRDye 680RD goat anti-rabbit IgG, respectively. Nucleolin and tubulin were used to demonstrate the purity of nuclear and cytoplasmic fractions.

In cells cotransfected with US3 and wild-type VP8, the cytoplasmic VP8 was dispersed (Fig. 4A), which was different from the juxtanuclear aggregation of VP8 shown at the late phase of BoHV-1 infection (Fig. 2B), which suggests that after leaving the nucleus, VP8 is translocated to the Golgi apparatus during infection. However, US3-mediated phosphorylation was not sufficient for VP8 to localize in the Golgi of transfected cells, indicating that the cytoplasmic VP8 might require an unidentified viral protein or stimulus to target the Golgi.

To demonstrate that the nuclear export of VP8 is enhanced with increasing amounts of US3, FLAG-VP8 was first transfected into EBTr cells and US3-hemagglutinin (HA) was subsequently transfected at different time points into the cells expressing FLAG-VP8 (Fig. 5). Prior to transfection with US3-HA, the average readout at the 633-nm channel was 6.9 procedure defined units (PDU), which was thus considered the baseline for US3 expression. From 0 to 15 h after transfection with pUS3-HA, the intensity of nuclear US3 increased to about 130 PDU. With the increase of US3, cytoplasmic VP8 gradually increased from 22 PDU to 120 PDU, while the intensity of nuclear VP8 did not significantly change, confirming that US3 was critical for VP8 to localize in the cytoplasm.

FIG 5.

The amount of cytoplasmic FLAG-VP8 increases with the expression level of US3-HA. (A) The intensities of cytoplasmic and nuclear VP8. (B) The intensity of US3 in the nucleus. EBTr cells were transfected with pFLAG-VP8 and incubated for 12 h. This allowed VP8 to be expressed and localized to the nucleus. Subsequently, cells were transfected with pUS3-HA at 12 hpi, 17 hpi, or 22 hpi after transfection with pFLAG-VP8. All samples were fixed at 27 h for immunofluorescent staining, and the cell images were analyzed with the software Leica Application Suite X for protein quantification. VP8 was detected with monoclonal anti-VP8 antibody followed by Alexa Fluor 488-conjugated goat anti-mouse IgG. US3 was identified with polyclonal anti-US3 antibody followed by Alexa Fluor 633-conjugated goat anti-rabbit IgG. The intensities of fluorescence pixels at wavelengths of 488 nm and 633 nm were captured to indicate the quantity of VP8 and US3, respectively. The mean PDU are shown in the bar graphs. Error bars represent the SDs. The statistical significance is shown as follows: *, 0.01 < P ≤ 0.05; **, P ≤ 0.01.

To determine whether phosphorylation by CK2 (8) regulates the nuclear localization of VP8, the cellular localization of VP8-M65-107 (8) was compared with that of wild-type VP8 in EBTr cells (Fig. 6A). VP8-M65-107 is not phosphorylated by CK2 because the residues critical for CK2-mediated phosphorylation (T⁶⁵, S⁶⁶, S⁷⁹, S⁸⁰, S⁸², S⁸⁸, and T¹⁰⁷) are mutated (8). Wild-type VP8 and VP8-M65-107 both localized in the nuclei of transfected cells, implying that CK2-mediated phosphorylation did not change the nuclear localization of VP8. In addition, the cellular localization of wild-type VP8 and VP8-M65-107 was examined in cells overexpressing CK2α-HA (Fig. 6B). The two versions of VP8 were localized in the nucleus in the presence and in the absence of CK2α-HA, confirming that CK2 is not involved in VP8 translocation.

FIG 6.

Phosphorylation by CK2 does not change the nuclear localization of VP8. (A) Mutating CK2-phosphorylated residues in VP8 does not affect the nuclear localization of VP8. EBTr cells were transfected with pFLAG-VP8 or pFLAG-VP8-M65-107 and fixed for immunofluorescent staining. (B) Overexpression of CK2α does not alter the VP8 nuclear localization. Plasmid pFLAG-VP8 or pFLAG-VP8-M65-107 was cotransfected with pCK2α-HA or individually transfected into COS-7 cells. The cells were fixed for immunofluorescent staining. VP8 was identified with polyclonal anti-VP8 antibody followed by Alexa Fluor 488-conjugated goat anti-rabbit IgG. CK2α-HA was identified with monoclonal anti-HA antibody followed by Alexa Fluor 633-conjugated goat anti-mouse IgG. The cell nuclei were identified with DAPI.

Interaction between importin β and VP8 is not affected by US3-mediated phosphorylation.

In the amino acid sequences of VP8 from strains 108 and Cooper (Fig. 7A), NLS1 (10, 13) is immediately followed by a phosphorylated residue S¹⁶ (8). Between the NLS1 and NLS2 (10, 11, 13) there is the second phosphorylation site at S³² (8). Alignment of amino acid sequences showed that the two US3 phosphorylation residues and NLS2 are conserved among two strains of BoHV-1. NLS1 is partially conserved with a 2-amino-acid variation in the Cooper strain compared to strain 108. The close locations of phosphorylated residues and NLSs suggested that phosphorylation by US3 may affect the function of the NLSs, rendering VP8 unable to enter the nucleus. To test this hypothesis, the impact of phosphorylation on the interaction between YFP-VP8 and glutathione S-transferase (GST)-importins was tested with pulldown experiments (Fig. 7B). The results showed that both YFP-VP8 and YFP-Mut-VP8 (8), which is not phosphorylated, interacted with GST-importin β. This indicates that importin β mediates the nuclear localization of VP8 and that phosphorylation at S¹⁶ and S³² does not influence this interaction.

FIG 7.

Phosphorylation does not affect the interaction between VP8 and importin β. (A) NLSs and US3-phosphorylated residues in partial sequences of VP8. The protein sequences of VP8 from strains 108 and Cooper were obtained from UniProKB. NLS1 and NLS2 are underlined. US3-phosphorylated residues are labeled with “*.” Gray color indicates conserved regions and white color indicates mutations or deletions. (B) YFP-VP8 and YFP-Mut-VP8 are associated with importin β. GST-fused importin proteins were expressed in plasmid-transformed BL21 cells. YFP-VP8 and YFP-Mut-VP8 were expressed by infecting MDBK cells with BoHV-1-YVP8 and BoHV-1-YmVP8, respectively. GST-fused proteins were detected with mouse anti-GST antibody followed by IRDye 800CW goat anti-mouse IgG. YFP-VP8 and YFP-Mut-VP8 were detected with rabbit anti-VP8 antibody followed by IRDye 680RD goat anti-rabbit IgG. After the GST pulldown, the presence of YFP-VP8 and YFP-Mut-VP8 was confirmed by Western blotting using the above-described antibodies.

Nuclear export of VP8 is sensitive to leptomycin B.

To determine whether the CRM1 pathway plays a role in the nuclear export of VP8 during BoHV-1 infection, a CRM1-specific inhibitor, leptomycin B (LMB), was used to treat BoHV-1-infected MDBK cells. Cells infected with wild-type BoHV-1 were cultured with or without LMB (Fig. 8). At 4 hpi VP8 was in the nucleus, while at 6 hpi the distribution of VP8 was different between the two treatments. VP8 formed perinuclear aggregates in the cell cytoplasm without LMB. In contrast, in LMB-treated cells VP8 remained in the nucleus and did not accumulate in the cytoplasm. When cells were cultured for 8 h without LMB the majority of the cells showed cytoplasmic accumulation of VP8, whereas cells with LMB displayed occasional cytoplasmic accumulation of VP8. These results showed that the nuclear export of VP8 was reduced and/or delayed by the CRM1-specific inhibitor but was not completely inhibited.

FIG 8.

The nuclear export of VP8 is sensitive to LMB. MDBK cells were infected with BoHV-1. Culture medium supplemented or not with LMB (20 nM) was applied to cells at 1 hpi. Medium with or without LMB was renewed every 2 h. Cells were fixed at 4, 6, and 8 hpi and incubated with monoclonal anti-VP8 antibody followed by Alexa Fluor 488-conjugated goat anti-mouse IgG. The cell nuclei were identified with DAPI.

VP8 is colocalized with cis-Golgi cisterna proteins.

The location of the Golgi cisternae was determined with antibodies specific for two Golgi proteins, formimidoyltransferase cyclodeaminase (FTCD) and golgin subfamily B member 1 (GOLGB1) (Fig. 9). FTCD has been found to associate with the Golgi membranes of the cis-cisternae and medial cisternae (18–20) and is involved in the conversion of l-histidine to l-glutamate (20, 21). GOLGB1 is a transmembrane Golgi protein (22) that mainly associates with the Golgi apparatus (23), with preference for the cis-Golgi (24). In mock-infected MDBK cells, FTCD and GOLGB1 showed almost identical localization in the Golgi area (Fig. 9, top image). This was consistent with previous findings that both proteins are in the cis-cisternae and the stacks of the Golgi (18–20, 23, 24). Thus, GOLGB1 was used as a marker protein to indicate the cis-Golgi and the Golgi stacks in BoHV-1-infected MDBK cells (Fig. 9, lower images). At 6 h 40 min postinfection, VP8 appeared in the cytoplasm, but most VP8 was localized in areas different from GOLGB1. At 7 h 20 min postinfection, the majority of VP8 was colocalized with GOLGB1 in the cytoplasm. The colocalization of VP8 and GOLGB1 persisted until 8 h after infection.

FIG 9.

Colocalization of VP8 and a cis-Golgi marker protein. MDBK cells mock infected or infected with wild-type BoHV-1 were fixed for immunofluorescent staining. Mock-infected cells shown in the top image were identified with mouse anti-FTCD and rabbit anti-GOLGB1 antibodies. BoHV-1-infected cells in the lower images were identified with mouse anti-VP8 and rabbit anti-GOLGB1 antibodies. Secondary antibodies were Alexa Fluor 488-conjugated goat anti-mouse IgG and Alexa Fluor 633-conjugated goat anti-rabbit IgG, respectively. The cell nuclei were identified with DAPI. Selected areas of the pictures in the first row were zoomed in and are shown on the right side.

During BoHV-1 infection, the Golgi apparatus went through obvious morphological changes. The cis-Golgi and the Golgi stacks illustrated with GOLGB1 were reshaped as a cluster of vesicular structures. VP8 gradually accumulated and overlapped with these structures. However, in mock-infected cells, the Golgi structures labeled with FTCD and GOLGB1 were multilayer sheets closely adjacent to the cell nucleus (Fig. 9).

VP8 is not present in the trans-Golgi network.

To determine whether VP8 was associated with the trans-Golgi network (TGN), brefeldin A (BFA) was used to distinguish the TGN from the cis-Golgi and medial Golgi. It is known that BFA causes redistribution of membranes and proteins of the Golgi cisternae, while the TGN responds differently to this compound and its perinuclear structure is not disrupted (25–27). To confirm this effect, FTCD and GOLGB1 were used to demonstrate the Golgi cisternae, including the cis-Golgi network (CGN). trans-Golgi network integral membrane protein 2 (TGOLN2), a protein that associates with the TGN to participate in the transport between the TGN and the cell surface (28), was used to indicate the TGN. The experiment was performed in Henrietta Lacks’ immortal (HeLa) cells, a commonly used human cell line, because antibodies against TGOLN2 are specific for the human TGOLN2 and do not cross-react in bovine cells. In nontreated HeLa cells, FTCD and TGOLN2 formed clear juxtanuclear aggregates (Fig. 10). In BFA-treated cells, FTCD was scattered at 2 h posttreatment, whereas TGOLN2 aggregates were reshaped but not dispersed. With BFA, the TGOLN2 aggregates persisted until 8 h posttreatment. Moreover, in cells that were incubated with antibodies against FTCD and GOLGB1, the two cis-Golgi proteins were scattered by BFA at 2 h, confirming that the cis-Golgi and medial Golgi but not the TGN were redistributed by BFA.

FIG 10.

BFA disperses the cis-Golgi proteins but not the TGN protein. HeLa cells were treated with BFA or DMSO. The BFA was renewed every 2 h. cis-Golgi proteins FTCD and GOLGB1 were detected with mouse anti-FTCD and rabbit anti-GOLGB1 antibodies. TGN protein TGOLN2 was detected with rabbit anti-TGOLN2 antibody. Alexa Fluor 488-conjugated goat anti-mouse IgG and Alexa Fluor 633-conjugated goat anti-rabbit IgG were used as secondary antibodies. The cell nuclei were identified with DAPI.

BoHV-1-infected cells were treated with BFA to determine its impact on VP8 distribution (Fig. 11). MDBK cells were infected with BoHV-1, followed by BFA treatment at 1 hpi. In the presence of BFA, nuclear VP8 appeared at 4 hpi. From 4 to 8 hpi the infected cells contained increasing amounts of cytoplasmic and nuclear VP8 with diffuse distribution (Fig. 11A). In the dimethyl sulfoxide (DMSO)-treated cells, cytoplasmic VP8 gradually accumulated in the juxtanuclear areas from 6 to 8 hpi. Thus, the cytoplasmic distribution of VP8 in DMSO-treated cells was profoundly different from that of VP8 in BFA-treated cells. A comparable impact of BFA on the Golgi localization of VP8 was observed in EBTr cells (data not shown). To demonstrate dispersal of Golgi-localized VP8, MDBK cells infected with BoHV-1-YVP8 were treated with BFA at 7 hpi, when VP8 was accumulated in the Golgi apparatus. The cells were observed with live-cell imaging confocal microscopy at 7 hpi for 20 min. Five minutes after BFA was applied, the amount of Golgi-localized YFP-VP8 started to decrease, and the cytoplasmic YFP-VP8 aggregates were dispersed within 20 min of BFA treatment.

FIG 11.

The Golgi accumulation of VP8 is sensitive to BFA. (A) MDBK cells infected with BoHV-1 were treated with BFA or DMSO at 1 hpi. Cells were fixed and incubated with monoclonal anti-VP8 antibody followed by Alexa Fluor 488-conjugated goat anti-mouse IgG. The cell nuclei were identified with DAPI. (B) MDBK cells infected with BoHV-1-YVP8 were treated with BFA at 7 hpi. Live cells showing YFP-VP8 were observed from 7 hpi for 20 min.

The localization of VP8 in the cis-Golgi was confirmed through highlighting the cis-Golgi and TGN with GOLGB1 and adaptor-related protein complex 1 γ-1 subunit (AP1G1), a TGN-associated protein involved in TGN-endosome trafficking (29), respectively. AP1G1 and GOLGB1 aggregated near the nucleus but did not colocalize in mock-infected MDBK cells, indicating that the two proteins associated with different compartments of the Golgi complex. In BFA-treated cells, GOLGB1 was dispersed but AP1G1 was not (Fig. 12A). In BoHV-1-infected cells, VP8 did not colocalize with AP1G1. The perinuclear aggregation of VP8 was dispersed by BFA treatment (Fig. 12B), indicating that VP8 associated with the same Golgi compartment as GOLGB1 but different from AP1G1.

FIG 12.

VP8 is not colocalized with TGN protein AP1G1. (A) GOLGB1 is dispersed by BFA treatment. MDBK cells were treated with BFA. AP1G1 was detected with AP1G1-specific monoclonal antibody followed by Alexa Fluor 488-conjugated goat anti-mouse IgG. GOLGB1 was detected with GOLGB1-specific polyclonal antibody followed by Alexa Fluor 633-conjugated goat anti-rabbit IgG. (B) VP8 is dispersed by BFA treatment. BoHV-1-infected MDBK cells were treated with BFA. VP8 was detected with VP8-specific polyclonal antibody followed by Alexa Fluor 488-conjugated goat anti-rabbit IgG.

VP8 causes nuclear lipid droplet formation.

In transfected EBTr cells, nuclear VP8 accumulated around ring-shaped subnuclear structures (Fig. 1 and 4). These VP8-induced subnuclear rings also appeared in VP8-transfected African green monkey fibroblast-like (COS-7) cells (8). To identify these subnuclear structures, COS-7 and human hepatoma (Huh-7) cells were incubated with Nile red (Fig. 13), a compound that is commonly used for the detection of intracellular neutral lipids (30, 31). Huh-7 cells, which naturally contain lipid droplets (LDs) (32), were used to confirm the lipid staining efficacy of Nile red. Numerous LDs of different sizes were observed in the cytoplasm of Huh-7 cells. The interior of the round structures within the nuclei of FLAG-VP8 (WT)-transfected cells were stained with Nile red, indicating that the expression of VP8 (WT) caused nuclear LD formation in transfected cells and that VP8 accumulated around the LDs, forming ring-shaped structures. There were no LDs in US3-HA- and FLAG-YFP-expressing cells, confirming that the LD formation was not due to viral protein expression or plasmid transfection in general. In addition, the formation of LDs in VP8-transfected cells did not depend on the phosphorylation of VP8, because cells expressing nonphosphorylated VP8 and Mut-VP8 also developed nuclear LDs.

FIG 13.

VP8 causes formation of nuclear lipid droplets in transiently transfected cells. COS-7 cells transfected with different plasmids and Huh-7 cells were fixed for immunostaining. VP8 (WT) was detected with VP8-specific monoclonal antibody followed by Alexa Fluor 488-conjugated goat anti-mouse IgG. US3 was detected with US3-specific polyclonal antibody followed by Alexa Fluor 488-conjugated goat anti-rabbit IgG. Mut-VP8 was detected with VP8-specific polyclonal antibody followed by Alexa Fluor 488-conjugated goat anti-rabbit IgG. Cells were then incubated with Nile red and DAPI.

To determine the mechanism of LD formation in BoHV-1-infected cells, MDBK cells were infected with wild-type BoHV-1, BoHV-1-YmVP8, and ΔUS3-BoHV-1 for 8 h (Fig. 14). In cells infected with wild-type BoHV-1, VP8 was mainly in the cytoplasm and there was no LD in the nucleus, while BoHV-1-YmVP8- and ΔUS3-BoHV-1-infected cells contained VP8 in the nucleus and developed nuclear LDs. These results indicate that without phosphorylation by US3 the majority of VP8 remained in the nucleus, leading to the LD formation, while US3-phosphorylated VP8 left the nucleus at late stages of infection. Consequently, LDs were not detected in the nucleus.

FIG 14.

Nuclear lipid droplets are inversely correlated with phosphorylation of VP8 in infected cells. MDBK cells were infected with Wt BoHV-1, BoHV-1-YmVP8, or ΔUS3-BoHV-1 and fixed for immunostaining. VP8 (WT) and Mut-VP8 were detected with VP8-specific polyclonal antibody followed by Alexa Fluor 488-conjugated goat anti-rabbit IgG. US3 was detected with US3-specific polyclonal antibody followed by Alexa Fluor 488-conjugated goat anti-rabbit IgG. Cells were then incubated with Nile red and DAPI.

To verify the impact of VP8 expression on the neutral lipid level of cells, cellular total cholesterol contents were compared between pFLAG-VP8- and pUS3-HA-transfected COS-7 cells (Fig. 15). The cholesterol level of pFLAG-VP8-transfected cells was about 1.75-fold higher than that in pUS3-HA-transfected cells (Fig. 15A), indicating that VP8 expression causes accumulation of cholesterol in cells. To confirm that VP8 increases the cholesterol level during BoHV-1 infection, total cholesterol levels were determined in MDBK cells infected with wild-type BoHV-1 or ΔU_L47-BoHV-1 (Fig. 15B). The total cholesterol level was significantly higher in wild-type BoHV-1-infected cells than in mock-infected and ΔU_L47-BoHV-1-infected cells.

FIG 15.

Cellular total cholesterol is increased in the presence of VP8. (A) Cellular total cholesterol is increased in VP8-transfected cells. COS-7 cells transfected with pFLAG-VP8 or pUS3-HA were harvested at 12 hpt. Cellular total cholesterol was quantified using a Cholesterol/Cholesteryl Ester Quantitation Assay kit. (B) Cellular total cholesterol is increased by BoHV-1 infection. MDBK cells were mock infected or infected with wild-type BoHV-1 or ΔU_L47-BoHV-1 at an MOI of 0.1 for 24 h. Cellular total cholesterol in MDBK cells was quantified with the above-described assay kit. **, P ≤ 0.01.

DISCUSSION

In this study, the regulation of the cellular translocation of VP8 during BoHV-1 infection was investigated. Early during infection VP8 was localized in the nucleus. VP8 contains two functional NLSs on the extreme N terminus (10) directing the protein to translocate into the nucleus early during infection to promote early events of virus replication, for instance, viral DNA encapsidation (12), and to redistribute ND10 (8). Subsequently, VP8 was translocated to the cytoplasm and accumulated in the Golgi apparatus. This might be a process of recycling VP8 from the nucleus to the cytoplasm so as to effectively use the protein and/or allow virion incorporation. US3 was essential and sufficient for the cytoplasmic localization of VP8 during the late stages of infection. VP8 remained in the cytoplasm, most likely because US3-mediated phosphorylation disturbed the nuclear import of VP8 or facilitated its nuclear export. However, the nuclear localization of VP8 did not require and was not affected by CK2-mediated phosphorylation (Fig. 6A). Overexpression of CK2α did not alter the nuclear localization of VP8 in transfected cells (Fig. 6B). This agrees with the fact that cellular kinase CK2 is universally expressed in almost every subcellular structure; therefore, it might be available to phosphorylate VP8 inside or outside the nucleus.

As a key factor that controls the cytoplasmic localization of VP8, US3-mediated phosphorylation correlated with the cytoplasmic intensity of VP8 (Fig. 3 and 4). Cytoplasmic VP8 gradually increased with the progression of US3 expression. However, the intensity of nuclear VP8 did not decrease with the increase of cytoplasmic VP8, which may be related to continuous synthesis of VP8 followed by import into the nuclei of transfected cells. During BoHV-1 infection, a regulatory system might exist to balance the nuclear import and export of VP8 at the right time. In VP8-transfected cells, VP8 mostly localizes in the nucleus; however, it translocates between two nuclei within a fused cell (10, 13), indicating that the nuclear import and export are active simultaneously and that the cytoplasmic VP8 can reenter the nucleus. Early during infection, VP8 might be required in the nuclear events. To ensure that the majority of VP8 is in the nucleus, the nuclear import should be more efficient than the export. Later during infection, a large amount of VP8 might be required in the cytoplasmic events, especially in the Golgi apparatus. Thus, the cytoplasmic VP8 should be stopped from reentering the nucleus by inactivating the nuclear import through US3. This function seems to be exclusive to US3. It does not appear to need an intermediate kinase to phosphorylate VP8, because US3 and CK2 are the only kinases that phosphorylate VP8 in BoHV-1-infected cells (12) and CK2 does not affect the nuclear localization of VP8.

The cytoplasmic location of VP8 in the presence of US3 could have two reasons. The function of the NLSs might be disrupted, leaving VP8 to be exported by NESs but unable to reenter the nucleus. The close localization of the NLSs and US3-phosphorylated residues (S¹⁶ and S³²) (8) in VP8 led to the hypothesis that phosphorylation at S¹⁶ and S³² blocks the NLS from binding to importins. To identify the importin protein that leads VP8 into the nucleus, the interaction between YFP-VP8 and a group of importin proteins (importin α1, importin α3, importin α5, importin α7, and importin β) was examined. The results showed that VP8 interacts with importin β but not with the other importins. Importin β-dependent nuclear import has been reported for human immunodeficiency virus type 1 (HIV-1) regulatory proteins Tat and Rev (33). Because both VP8 and Mut-VP8, a nonphosphorylated VP8, interacted with importin β, phosphorylation through US3 is less likely to affect the NLSs of VP8. However, this does not exclude the possibility that US3-mediated phosphorylation affects the efficiency of VP8 release from importin in the interior of the nuclear membrane, which is another important decisive factor for nuclear localization of proteins (34). The second possible reason is that the phosphorylation by US3 increases the potency of NESs, which leads to cytoplasmic localization of VP8. An example that phosphorylation enhances the efficacy of NESs is Pho4, a transcription factor (35). There are at least two NESs in VP8. A leucine-rich peptide, L⁴⁸5SAYLTLFVAL⁴⁹⁵ (NES1), of VP8 is a classic CRM1-dependent signal, the activity of which is sensitive to LMB (10, 11, 13). Another NES that is CRM1 independent exists in the region between residues 95 and 123 (NES2) (13). LMB treatment reduced and/or delayed nuclear export of this protein during BoHV-1 infection (Fig. 8), which might be due to inhibition of NES1 by LMB. However, LMB did not completely block the cytoplasmic localization of VP8, which is consistent with the characteristic of NES2. During the late phase of infection, both NESs might be functional to ensure the nuclear export of VP8. Mutating both NESs does not completely block VP8 translocation from one nucleus to another within the same cell generated by interspecies heterokaryons (10, 13), suggesting another unidentified NES in VP8.

There are minor variations in the sequences of VP8 and US3 between the 108 and Cooper strains. However, these differences do not affect the mechanism and function of US3-mediated phosphorylation in VP8. The conserved catalytic loop and ATP-binding pockets in US3, and the US3 target sites and NLSs in VP8, are not affected by the mutations. Thus, US3-regulated cellular localization of VP8 was expected to be conserved among the tested strains of BoHV-1. Consistently, the patterns of nuclear-cytoplasmic translocation of VP8 were identical in strains 108 and Cooper.

After leaving the nucleus, VP8 gradually migrated from the cytoplasm toward the Golgi and colocalized with a cis-Golgi protein. Although the cytoplasmic localization of VP8 requires US3-mediated phosphorylation, the subsequent Golgi-localization is not directly related to US3, suggesting that VP8 needs further modifications by or interactions with other viral factors in the cytoplasm to accumulate in the Golgi.

The Golgi apparatus is an important site for cytoplasmic maturation of herpesviruses. This organelle is composed of several continuous compartments, including the CGN, cis-cisternae, medial cisternae, trans-cisternae, and TGN (36). They play different yet related roles in protein sorting, modification, transport, and secretion (37, 38). Therefore, the localization of VP8 in the Golgi compartments might reveal a potential function of the Golgi-localized VP8. We demonstrated colocalization of VP8 and a cis-Golgi marker protein, indicating that VP8 is translocated to the cis-Golgi but not to the TGN. BFA is an antifungal drug that causes rapid and extensive disorganization of the cisternal Golgi in mammalian cells by redistributing the membrane and enzymes from the Golgi cisternae to the endoplasmic reticulum (ER) (25, 27). However, the response of the TGN to BFA is dissimilar to that of the Golgi cisternae (27). This response was confirmed in HeLa and MDBK cells, in which TGN protein TGOLN2 and AP1G1 were accumulated in the juxtanuclear space when cultured with BFA, whereas the cis-Golgi proteins FTCD and GOLGB1 were dispersed. BFA blocked the Golgi localization of VP8 and disrupted the Golgi-localized VP8. A similar impact of BFA was found in the distribution of the cis-Golgi proteins but not in the TGN proteins, indicating that VP8 is associated with the cisternal Golgi rather than the TGN. In infected cells, VP8 did not colocalize with AP1G1, confirming that VP8 was not in the TGN. As a result, it is less likely to be involved in viral particle transport through the TGN. We have previously shown that inhibiting VP8 from entering the cytoplasm significantly reduced the amount of virion-incorporated VP8 (12). This is consistent with VP8 being localized in the cis-Golgi, which is an important site for protein incorporation into virions.

Nuclear LDs were formed in VP8-transfected cells (Fig. 13), suggesting that VP8 caused an abnormal lipid metabolism in cells. In general, LDs can be found in the cytoplasm (39) and the nucleus (40, 41), serving as lipid storage and being involved in cellular lipid metabolism (42). Accordingly, the extra lipid produced in VP8-transfected cells might be stored as nuclear LDs. This ability was unique to VP8, because nuclear LDs were not found in US3-transfected or YFP-transfected cells. To induce the nuclear LDs VP8 did not have to be phosphorylated but needed to remain in the nucleus. Wild-type VP8 and Mut-VP8 both were in the nucleus of transfected cells and caused nuclear LD formation. In late stages of BoHV-1 infection, nonphosphorylated VP8 was accumulated in the nucleus and had the capacity to induce LD formation (Fig. 14), while US3-phosphorylated VP8 was cytoplasmic and consequently, no LDs were formed in the nucleus.

Cholesterol is one of the lipids that accumulates in herpesvirus-infected cells (43). It is required for HHV-1 entry (44) and for the replicative cycle at later stages after entry (45). Therefore, the cellular total cholesterol was examined in BoHV-1-infected cells. The cellular total cholesterol was not significantly increased in ΔU_L47-BoHV-1-infected cells, confirming that the lipid increase was caused by VP8 expression. The cellular total cholesterol was increased in wild-type BoHV-1-infected cells, indicating that the cellular lipid metabolism was changed by infection. However, the increased lipid was not accumulated as LDs, suggesting that the lipid was involved in viral replication and presumably was used for virus assembly and transport. Such a function has been found for other herpesviruses. One example is HHV-1 infection, which induces de novo synthesis of phospholipid to maintain cell membrane integrity and to supply membrane constituents for envelopment of capsids and transport vacuoles (46). Another example is human herpesvirus 8 (HHV-8). Lipid accumulation has been reported for HHV-8-infected cells during the lytic phase of infection (47). HHV-8 infection increases the cellular triglycerides that are stored in the cytoplasmic LDs, which is related to the abnormal viral requirements for lipids (47). The formation of nuclear LDs in BoHV-1-YmVP8- and ΔUS3-BoHV-1-infected cells confirmed that infection increased lipid formation and suggested that the consumption of lipid was related to US3-phosphorylated VP8.

Overall, our data show that VP8 translocates from the nucleus to the cytoplasm when BoHV-1 infection progresses to the late stages and that US3-mediated phosphorylation is required for VP8 to exit the nucleus and/or remain in the cytoplasm. During early stages of infection, the majority of VP8 is present in the cell nucleus to promote nuclear events for viral replication, such as redistributing promyelocytic leukemia (PML) protein (8) and promoting viral DNA encapsidation (12). Subsequently, the NLSs appear to be impaired or the NESs enabled by US3-mediated phosphorylation, resulting in VP8 accumulation in the cytoplasm. In the cytoplasm, VP8 is directed to the cis-Golgi, for which other factors are required, presumably to be incorporated into viral particles. The total cholesterol level was increased in BoHV-1-infected cells but not in ΔU_L47-BoHV-1-infected cells, indicating that VP8 plays a role in altering the cellular lipid metabolism during infection.

MATERIALS AND METHODS

Viruses and cell lines.

MDBK cells, Huh-7 cells, COS-7 cells, EBTr cells, fetal bovine testicular (FBT) cells, and HeLa cells were cultured in Eagle’s minimum essential medium (MEM) (Gibco/Thermo Fisher Scientific, Waltham, MA) supplemented with 10 mM HEPES (Gibco), 1% nonessential amino acids (Gibco), and 10% fetal bovine serum (FBS; Gibco).

BoHV-1 strain 108 was the parental virus of BoHV-1-YVP8 (which expresses YFP-VP8), BoHV-1-YmVP8 (which expresses Mut-VP8), and ΔU_L47-BoHV-1 (which does not express VP8) (4, 12). Production of all viral stocks was carried out in MDBK cells as previously described (8). Briefly, virus infections were accomplished by incubating 85% to 90% confluent cell monolayers in 150-cm² flasks with 10 ml of diluted virus at 37°C, which was replaced with 15 ml of MEM with 2% FBS at 1 hpi. Viruses were collected when cytopathic effect was well developed. The virus titers were determined by plaque titration on MDBK cells in 24-well plates overlaid with 8% UltraPure low-melting-point agarose (Invitrogen/Thermo Fisher Scientific) in MEM.

To generate a US3 deletion mutant, a ΔUS3 genotype of BoHV-1 in a bacterial artificial chromosome (BAC) was created by homologous recombination between pCooper BAC (full-length genome of BoHV-1 strain Cooper in BAC plasmid) and a DNA fragment containing a kanamycin resistance cassette flanked by upstream and downstream regions of the US3 gene in Escherichia coli DH10B cells (48, 49; T. Donovan and S. van Drunen Littel-van den Hurk, unpublished data). Colonies containing pΔUS3-BoHV-1 BAC were selected for by kanamycin and chloramphenicol resistance and the deletion was verified by DNA sequencing. The US3 revertant (pRUS3-BoHV-1) was rescued using en passant mutagenesis (50). The revertant BAC plasmid was verified by PCR, DNA sequencing, and restriction digestion. Viruses were generated by transfecting pΔUS3-BoHV-1 and pRUS3-BoHV-1 into FBT cells.

Antibodies and plasmids.

The antibodies included monoclonal antibodies specific for VP8 (clone 1G4 2G2) (5), FLAG (Sigma-Aldrich, St. Louis, MO), AP1G1 (Sigma-Aldrich), and FTCD (Sigma-Aldrich) and polyclonal antibodies specific for VP8 (5), US3 (17), GOLGB1 (Sigma-Aldrich), and TGOLN2 (Thermo Fisher Scientific). Alexa Fluor 488-conjugated goat anti-mouse IgG and Alexa Fluor 633-conjugated goat anti-rabbit IgG (Thermo Fisher Scientific) were used for immunofluorescent staining. IRDye 680RD goat anti-rabbit IgG and IRDye 800CW goat anti-mouse IgG (Li-Cor Biosciences, Lincoln, NE) were used for Western blotting.

Plasmids pFLAG-VP8 and pUS3-HA were previously generated by amplification from the cDNA of the BoHV-1 108 strain (9). Plasmids pFLAG-VP8-S16A and pFLAG-VP8-M65-107 were modified from pFLAG-VP8 by site-directed mutagenesis (8). DNA sequencing was performed by the NRC-Plant Biotechnology Institute (Saskatoon, SK, Canada). Plasmids containing GST-fused importin α1, α3, α 5, α 7, or β were a gift from M. Kӧhler (51). The plasmid pGST-transportin-SR2 (TRN-SR2) was provided by Woan-Yuh Tarn (52). Plasmid pCK2α-HA was purchased from Addgene (Watertown, MA).

Determination of cellular total cholesterol.

Cells cultured in 100-mm dishes were collected to determine cellular total cholesterol levels by using a cholesterol/cholesteryl ester quantitation assay kit (Abcam, Toronto, ON, Canada) according to the manufacturer’s manual. Briefly, cellular lipids were extracted by suspending and homogenizing cells in chloroform-isopropanol-nonyl phenoxypolyethoxylethanol (NP-40) (7:11:0.1). After homogenization, cell debris was removed by centrifugation, and supernatants were air dried. Trace organic solvent was removed in a SpeedVac concentrator. Dried cellular lipids were dissolved in assay buffer provided in the kit. Reaction mixtures containing 50 μl of cellular lipid solution or cholesterol standard solution, 0.4 μl of cholesterol probe, 2 μl of cholesterol esterase, 2 μl of enzyme mix, and 45.6 μl of cholesterol assay buffer were prepared in a black 96-well plate with a clear bottom. After incubation at 37°C for 60 min, fluorescent output was measured at excitation and emission wavelengths of 531 and 595 nm, respectively. A standard curve was prepared using cholesterol standard solution, and the concentrations of cellular total cholesterol were calculated according to the standard curve. The final result was expressed as cholesterol content (in micrograms) per 10⁶ cells.

GST pulldown.

The GST pulldown experiment was performed as previously described (53). Cell lysates containing YFP-VP8 or YFP-Mut-VP8 were prepared by infecting MDBK cells with BoHV-1-YVP8 or BoHV-1-YmVP8 at a multiplicity of infection (MOI) of 5. Cells were collected at 7 hpt and lysed with radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1% NP-40, 1% deoxycholate, 0.1% SDS) . Plasmids containing GST-fused importin α1, α3, α5, α7, or β or TRN-SR2 were individually transformed into E. coli BL21 cells. Cells were then treated with isopropyl-β-d-thiogalactopyranoside (IPTG) (1 mM) for 3.5 h at 37°C to induce protein expression. Glutathione Sepharose 4B (GE Healthcare, Chicago, IL) was incubated with the bacterial cell lysates to collect the GST-fused proteins. The beads were then incubated with virus-infected MDBK cell lysates. Associated proteins were eluted with GST elution buffer (50 mM Tris HCl, 10 mM reduced glutathione, 400 mM KCl) at 4°C for 20 min. The proteins were separated by 10% SDS-PAGE and detected by Western blotting.

Immunofluorescent staining and quantification.

Cell cultures were prepared in permanox chamber slides (Thermo Fisher Scientific). After being washed with phosphate-buffered saline (PBS; 136.9 mM NaCl, 2.7 mM KCl, 7.0 mM Na₃PO₄, 0.9 mM K₃PO₄ [pH 7.4]), the cultures were fixed with 4% paraformaldehyde for 20 min and again washed with PBS. The fixed cells were permeabilized with 0.1% Triton X-100 in PBS for 30 min, washed with PBS, and blocked with 1% normal goat serum (Gibco) for 30 min. To identify the proteins of interest, the cells were incubated with primary antibodies for 2 h and then with secondary antibodies for 1 h. Cells were washed three times with PBS after each antibody incubation. All incubations and washes were performed at room temperature. To identify the nucleus, the slides were incubated with 4′,6-diamidino-2-phenylindole (DAPI; 0.5 μg/ml) for 10 min at 37°C. To identify lipids, cells were incubated with Nile red (10 μg/ml; MP Biomedicals, Santa Ana, CA). To identify the trans-Golgi, cells were treated with BFA (5 μg/ml; Biolegend, San Diego, CA) dissolved in DMSO or with DMSO (0.1%). Finally, the slides were mounted with ProLong gold antifade mountant (Thermo Fisher Scientific) prior to examination with a Leica SP8 confocal microscope (Leica Microsystems CMS GmbH, Mannheim, Hesse, Germany).

Three-color images were generated by sequential scanning with 488-nm, 633-nm, and 461-nm lasers. A computer program, Leica AF Lite (Leica Microsystems CMS GmbH), was used to process the images and to calculate the relative quantification of fluorescent intensities according to the manufacturer’s manual. After making selections of the nuclear and cytoplasmic areas, the mean intensity of each channel within the defined area was quantified by the software and shown as PDU.

Preparation of nuclear and cytoplasmic fractions.

Nuclear and cytoplasmic fractions were prepared with a Nuclei EZ prep nuclei isolation kit (Sigma-Aldrich) by following the manufacturer’s instructions. Briefly, cell cultures were harvested with trypsin and washed with cold PBS. After centrifugation at 500 × g for 5 min, the cell pellets were lysed with cold Nuclei EZ lysis buffer supplied with a protease inhibitor cocktail (Sigma-Aldrich) for 5 min. The cytoplasmic fraction in the supernatant and nuclear fraction in the pellet were separated by centrifugation at 500 × g for 5 min. The pellets were repeatedly lysed with cold Nuclei EZ lysis buffer until all cell membranes were disrupted. The nuclear fragments were lysed in RIPA buffer supplemented with a protease inhibitor cocktail.

Western blotting.

Cell pellets were lysed in RIPA buffer supplied with a protease inhibitor cocktail. The cell lysates were clarified by centrifugation at 13,000 × g for 10 min at 4°C. The clarified lysates were boiled in SDS-PAGE sample buffer for 5 min. Proteins were separated by SDS-PAGE. After separation, they were transferred to nitrocellulose membranes and incubated with primary antibodies. The membranes were then washed and incubated with IRDye 680RD/800CW-conjugated IgGs. The resulting membranes were scanned with an Odyssey CLx infrared imaging system (Li-Cor Biosciences).

Statistical analysis.

Data were analyzed with Microsoft Excel 2010. Standard deviations were calculated based on the entire population of each group and are shown in figures as error bars. A two-tailed Student t test with two-sample unequal variances was used to determine the statistical differences between two independent treatments. If the P value was >0.01 and ≤0.05, differences between two treatments were considered statistically significant; if the P value was ≤0.01, differences were considered statistically highly significant.