Cyclin-Dependent Kinase 1/Cyclin B1 Phosphorylates Varicella-Zoster Virus IE62 and Is Incorporated into Virions

Abstract

Varicella-zoster virus (VZV), an alphaherpesvirus restricted to humans, infects differentiated cells in vivo, including T lymphocytes, keratinocytes, and neurons, and spreads rapidly in confluent cultured dermal fibroblasts (HFFs). In VZV-infected HFFs, atypical expression of cyclins D3 and B1 occurs along with the induction of cyclin-dependent kinase (CDK) activity. A specific CDK1 inhibitor blocked VZV spread, indicating an important function for this cellular kinase in VZV replication. CDK activity assays of infected cells revealed a large viral phosphoprotein that was identified as being the major immediate-early transactivator, IE62. Since IE62 colocalized with CDK1/cyclin B1 by confocal microscopy, we investigated whether this cellular kinase complex interacts with IE62. Using recombinant fragments of IE62 spanning the entire amino acid sequence, we found that purified CDK1/cyclin B1 phosphorylated IE62 at residues T10, S245, and T680 in vitro. Immunoprecipitation of cyclin B1 from VZV-infected HFFs indicated that IE62 was included in the complex within infected cells. The full-length IE62 protein, obtained by immunoprecipitation from infected cells, was also phosphorylated by purified CDK1/cyclin B1. Based on IE62/CDK1/cyclin B1 colocalization near viral assembly regions, we hypothesized that these cellular proteins could be incorporated into VZV virions with IE62. Purified virions were analyzed by immunoblotting for the presence of CDK1 and cyclin B1, and active CDK1 and cyclin B1 were present in the VZV tegument with IE62 and were sensitive to detergent treatment. Thus, IE62 is a substrate for CDK1/cyclin B1, and virions could deliver the active cellular kinase to nondividing cells that normally do not express it.

Varicella-zoster virus (VZV) (human herpesvirus type 3) infects epithelia, skin, T cells, and neurons and causes the human diseases chickenpox (varicella) and shingles (herpes zoster). The 125-kbp double-stranded viral DNA genome encodes 69 open reading frames (ORFs); this large proteome mediates essential viral functions and manipulates the cellular environment. The major immediate-early transactivator protein, IE62, is encoded by duplicated ORFs 62 and 71 and, as a tegument protein, is delivered to newly infected cell nuclei, where it initiates VZV replication by transactivating viral immediate-early and early genes (23, 25). While IE62 binds several viral proteins such as ORF4 (49), ORF9 (7), ORF47 (37), and ORF63 (31) and cellular transcription factors including upstream stimulatory factor, TATA-binding protein, and Sp1 (43-45), there is no information regarding interactions with cyclin-dependent kinases (CDKs). Roscovitine, a CDK inhibitor, blocks VZV replication in cell culture and in an ex vivo skin model, but it is not clear how CDK activity benefits the virus (51, 52).

VZV encodes two viral kinases, ORF47 and ORF66, both serine/threonine kinases that phosphorylate IE62 (10, 37). ORF66 phosphorylation of IE62 is responsible for the nuclear-cytoplasmic shuttling of this protein later in infection, while ORF47 phosphorylation has not been attributed a function (10, 21, 24, 26, 37). ORF66 and ORF47 truncation mutants are both viable in cultured cells, although ORF47 is essential in T cells in vivo (35). Other proteins, presumably cellular kinases, can thus compensate for the loss of ORF66 and ORF47 during infection of cultured cells (18, 47). The ORF47 protein has limited similarity to casein kinase II (CKII); ORF47 differs from CKII by the usage of GTP as a phosphate donor and inhibition by heparin (20). The catalytic and activation loops of ORF66 (gi:66866099) have 41% amino acid identity to CDK-like 5 (CDKL5) (gi:4507281) and 39% identity to CDK1 (gi:89161187), but ORF66 kinase is not inhibited by roscovitine (our unpublished observations).

Cellular kinases play an important role in the phosphorylation of VZV proteins. The IE63 protein is phosphorylated by CKI and CKII, resulting in nuclear localization for lytic or cytoplasmic retention in latent infection (3). When phosphorylated residues were mutated to alanine to block phosphorylation, the transfected IE63 protein shuttled from the nucleus to the cytoplasm, indicating a possible role for casein kinases in IE63 nuclear localization. When consensus CDK1 phosphorylation sites in IE63 were mutated to alanine, the protein became exclusively nuclear, whereas a mutation to glutamic acid resulted in a cytoplasmic localization (15). Glycoprotein E (gE) is phosphorylated by CKI and CKII in vitro, although no functional significance has been determined (13). The endodomain of glycoprotein I (gI) is phosphorylated by CDK1 in vitro (58).

IE62 is a constituent of the virion tegument, an amorphous layer of proteins between the viral capsid and envelope (23, 25). The mechanism of IE62 incorporation into the virion is unknown, but it is hypothesized to occur via interactions between proteins embedded within the envelope and tegument. The presence of cellular proteins within herpesvirus virions has been documented, such as CKII within human cytomegalovirus (HCMV) particles and β-actin in Kaposi's sarcoma-associated herpesvirus and pseudorabies virus (PRV) virions (2, 39), although a recent review stated that no virion proteomic studies of human alphaherpesviruses have been reported (39). Thus, little is known about the cellular components of VZV virions or how they are incorporated. It is possible that interactions with virion-associated viral proteins could assemble cellular proteins into the virion particle.

We have previously shown that roscovitine, an inhibitor of CDKs 1, 2, 5, 7, and 9, blocks VZV replication in cultured cells and a whole-skin explant model and halts IE62 translocation from the nucleus to the cytoplasm (51, 52). In addition, VZV induces specific cell cycle proteins in human fibroblasts, including CDK1, cyclin B1, cyclin D3, and cdc25C, without concurrent host DNA replication, indicating a shift of function for these proteins (28). Our previous finding that cyclin B1 and IE62 colocalized in the cytoplasm of infected cells led us to investigate the hypothesis that these proteins interact directly. Here, we demonstrate that IE62 is a substrate of CDK1/cyclin B1 in infected fibroblasts using purified proteins in vitro. Cyclin B1 and IE62 interact in infected cells, and using purified VZV virions, we demonstrate for the first time that active CDK1 and cyclin B1 are captured in the VZV tegument.

MATERIALS AND METHODS

Cells and virus.

Human foreskin fibroblasts (HFFs) and MeWo cells (human melanoma cell line) were cultivated as described previously (28). Strain rPOka of VZV was used for all infections (38). All media and supplements were purchased from Mediatech (Washington, DC), except for fetal bovine serum (Gemini Bio-Products, West Sacramento, CA).

Immunoprecipitations.

Samples were harvested in radioimmunoprecipitation assay buffer and immunoprecipitated as described previously (28). Bound proteins were eluted with reducing sample buffer. Where indicated, 50 μg/ml ethidium bromide was added to all immunoprecipitation buffers.

Fusion protein production and purification.

IE62 fragments were cloned into vector pMalCRI (New England Biolabs, Ipswich, MA), forming N-terminal fusions with maltose-binding protein (MBP) (22). Constructs were transformed into BL21-AI cells (Invitrogen, Carlsbad, CA). Colonies were selected using 100 μg/ml ampicillin and grown to an optical density at 600 nm of 0.4. A total of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) and 0.2% l-arabinose (Sigma) were added for protein expression at 32°C. Soluble proteins were collected using B-PER extraction reagent (Pierce Biotechnology) according to the manufacturer's directions. MBP-tagged proteins were purified using affinity chromatography over amylose beads (New England Biolabs). Fifteen fractions were collected and separated on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels, and protein content was analyzed using Coomassie blue staining or immunoblotting with an MBP antibody (New England Biolabs) (data not shown). Proteins were stored in 50% (vol/vol) glycerol (in distilled water).

Mutagenesis.

Plasmids were maintained in Escherichia coli DH5α cells (Invitrogen). DNA was purified (Qiagen) and mutagenized using the GeneTailor site-directed mutagenesis kit (Invitrogen). Primers (Sigma-Genosys) were designed according to the kit instructions, with serine or threonine residues mutated to alanines. DNA was methylated, and PCR was performed using Platinum Taq High Fidelity (Invitrogen) under the following conditions: 94°C for 2 min, denaturation at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 68°C for 7 min for 20 cycles, followed by 68°C for 10 min and storage at 4°C. DH5α bacteria were transformed with the PCR products, and colonies were selected using 100 μg/ml ampicillin. All mutants were confirmed by DNA sequencing (SUNY Upstate Core Facility).

Mutagenesis primers.

All primers read 5′ to 3′; mutations are in lowercase letters. The following primers were used: T10Aforward (CCGATGCAGCGCTCTgCACCCCAACGCGC), T10Areverse (AGAGCGCTGCATCGGCGGCGTATCCAT), S141Aforward (TCGGAACGAAGCTTGCTAGGgCTCCAAAACCACC), S141Areverse (CCTAGCAAGCTTCGTTCCGAGAGAGACTGT), S245Aforward (CCCGCTCAGGGAAAGgcaCCGAAGAAAAA), S245Areverse (CTTTCCCTGAGCGGGCCGTTGAGTTTTCT), T680Aforward (CACCGGATGATCGTTTACGAgctCCGCGCAAGCG), and T680Areverse (TCGTAAACGATCATCCGGTGGACACACAGA).

Immunoblotting.

Proteins were separated on 10% SDS-PAGE gels unless otherwise noted in the text, as described previously (28). Antibodies used include anti-IE62, anti-IE63, anti-ORF29, anti-ORF61, and anti-ORF4 rabbit sera; anti-IE62 monoclonal H6, anti-cyclin B1 (rabbit serum sc-752, monoclonal antibody [MAb sc-7393]), and anti-CDK1 (sc-54; Santa Cruz Biotechnology); anti-MBP (New England Biolabs); and monoclonal anti-gE 3B3, kindly provided by Charles Grose, University of Iowa. Alkaline phosphatase-conjugated anti-rabbit and -mouse antibodies were purchased from Jackson ImmunoResearch (West Grove, PA), and detection was performed by enhanced chemiluminescence (Pierce Biotechnology).

Microscopy.

Confluent HFFs were grown in six-well plates and inoculated with a 1:40 dilution of rPOka/HFFs for 2 h. Fresh medium containing 50 μM CDK1 inhibitor III (Calbiochem, San Diego, CA) or an equal volume of dimethyl sulfoxide (DMSO) was added, and the medium was refreshed every 24 h for 2 days. Cells were fixed in formalin and processed for immunohistochemistry by permeabilization with 0.2% Triton X-100 in phosphate-buffered saline, followed by staining with polyclonal human serum from a patient with recurrent zoster (graciously provided by Ann Arvin, Stanford University) and alkaline phosphatase-conjugated anti-human antibodies. Antibodies were detected with Fast Red staining and visualized using a Spot camera under a 4× objective. The microscope was calibrated using a stage micrometer, and plaque size was quantitated using the measurement function of the Spot program. A minimum of 25 plaques per treatment were analyzed, and statistically significant values were obtained following a Student's t test, with a P value of less than 0.05. In parallel, MeWo cells were seeded onto glass chamber slides, inoculated with a 1:50 dilution of rPOka/HFFs, and washed with fresh medium 2 h postinfection. Medium was replaced with 50 μM CDK1 inhibitor III or an equal volume of DMSO and refreshed 24 and 48 h postinfection. Cells were fixed in ice-cold methanol-acetone (1:1, vol/vol) and processed for immunofluorescence as described previously (28). Proteins visualized included IE62 and cyclin B1. In addition, immunofluorescent microscopy was performed on VZV-infected HFFs 48 h postinfection for cyclin B1, IE62, and TGN46 (Serotec) as described previously (28).

Kinase assays.

Immunoprecipitations or MBP-tagged IE62 fusion proteins were mixed with 10 ng recombinant CDK1/cyclin B1 (Upstate Biotechnology, Charlottesville, VA) and then incubated in kinase assay buffer (50 mM Tris [pH 7.5], 10 mM MgCl₂, 2.5 mM EGTA, 1 mM dithiothreitol, 0.1 mM ATP, and 1 μCi [γ-³²P]ATP) for 30 min at 37°C as described previously (28). The reaction was stopped with denaturing sample buffer, and proteins were separated on 10% SDS-PAGE gels. Signal was developed using a PhosphorImager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA). Virion kinase assays were performed in a modified kinase assay buffer (50 mM Tris [pH 7.6], 30 mM MgCl₂, 1 mM dithiothreitol, 0.1 mM ATP, and 1 μCi [γ-³²P]ATP) (25). Briefly, virion envelopes were solubilized with 0.5% Triton X-100 in water, followed by preincubation with either DMSO or 50 μM CDK1 inhibitor III (the 50% inhibitory concentration against CDK1/cyclin B1 is 28.8 μM) (Calbiochem, San Diego, CA) without ATP and then with the addition of 0.1 mM ATP and 1 μCi [γ-³²P]ATP. Proteins were separated, and phosphorylation was detected as described above.

VZV virion preparation and electron microscopy.

Twenty-four T175 flasks of confluent HFFs or 12 flasks of subconfluent MeWos were inoculated with a 1:4 dilution of ∼80%-cytopathic-effect rPOka in HFFs. At an ∼80% cytopathic effect (approximately 3 days postinfection), cells were scraped into spent medium and pelleted at 250 × g for 5 min. The pellets were resuspended in spent medium and Dounce homogenized until cell lysis occurred, but nuclei remained intact as determined by light microscopy. Nuclei were pelleted by centrifugation at 250 × g for 10 min and discarded; the resulting supernatant contained cytosol, organelles, and virion-containing vesicles and was centrifuged at 50,000 × g in a Beckman-Coulter L-90 ultracentrifuge using an SW-32 rotor for 2 h at 4°C. The resulting pellet was resuspended in 2 ml of medium, incubated at 4°C overnight, and then sonicated and centrifuged at 300 × g for 10 min twice to discard heavy materials (membranes and organelles). The supernatant was overlaid on a 5 to 15% Ficoll gradient prepared with essential modified Eagle's medium and ultracentrifuged at 24,000 × g in an SW-32 rotor, and the band corresponding to intact virions (two-thirds from the top of the gradient) was harvested using a 27-gauge needle and resuspended in a fourfold volume of medium. In a separate experiment, the Ficoll gradient was fractionated into 1-ml aliquots and diluted with 3× sample buffer for SDS-PAGE analysis. Virus was pelleted at 110,000 × g in an SW-40 rotor, resuspended in medium, and stored at −80°C. For negative-stain electron microscopy, virions were layered onto mica-coated carbon type A grids (Ted Pella, Redding, CA), washed once with water, and stained with 1% uranyl acetate for 1 min. Virions were dried onto the grids and viewed by transmission electron microscopy (Tecnai BioTWIN 12 microscope; FEI, Hillsboro, OR). Images were acquired using a Kodak ES4.0 Advantage Plus digital charge-coupled-device system 2K-by-2K camera (AMT, Danvers, MA).

PCR of DNA within purified virions.

DNA from infected HFF nuclei, purified virions, and a plasmid encoding full-length IE62 (pCMV62) was amplified using standard PCR conditions. DNA was liberated from nuclei and the virions by incubation at 95°C for 10 min. β-Actin primers were purchased from Stratagene (La Jolla, CA). PCR cycling conditions included an initial denaturation step at 94°C for 5 min, denaturation at 94°C for 30 s, annealing at 50°C for 30 s, extension at 72°C for 45 s repeated for 30 cycles, and a final extension step at 72°C for 5 min; primers used were as follows: IE62 forward primer TGTAATCCCGCTGGCCGAGGTCTT and IE62 reverse primer CTGAGCGGGCCGTTGAGTTTTCTG.

Virion treatments.

Purified virions were biochemically fractionated as described previously (57). Briefly, outer envelope proteins were digested with 50 μg trypsin (Sigma), or membranes were permeabilized with 0.5% Triton X-100 and 0.5% sodium deoxycholate, followed by acetone precipitation of soluble proteins. Tegument proteins were digested using a combination of detergent and trypsin treatments. Immunoblots were performed as described above.

RESULTS

CDK1 inhibition results in decreased VZV replication.

To determine the importance of active CDK1/cyclin B1 in VZV replication, infected cells were treated with a specific CDK1 inhibitor for 48 h. CDK inhibitor III treatment decreased VZV yield in HFFs 10-fold (data not shown). In addition, plaques were significantly smaller, shrinking from an average of 87 μm² to 18 μm², which consisted of singly infected cells (Fig. 1A). Immunofluorescence microscopy for cyclin B1 and IE62 showed that cyclin B1 was reduced in the presence of CDK1 inhibitor III in infected MeWos (Fig. 1B), the IE62 signal was dimmed, and syncytium formation was reduced. Typical MeWo syncytia consist of cell fusions encompassing 50 to 100 nuclei 2 days postinfection, as seen in the DMSO-treated panel. However, few cells showed cytopathic effects or syncytia in the presence of CDK1 inhibitor III.

FIG. 1.

CDK1 inhibition reduces VZV syncytium formation. (A) HFFs were grown in six-well plates infected with rPOka and treated with 50 μM CDK1 inhibitor III (CDK1 Inh.) or an equal volume of DMSO for 2 days. Cells were fixed and processed for viral glycoprotein immunohistochemistry. Plaque size was quantitated under a 4× objective using the Spot camera program. The asterisk indicates statistical significance (P < 0.01 by the Student t test [P = 3 × 10⁻¹²]). (B) MeWo cells were grown on chamber slides, infected with VZV rPOka, and simultaneously treated with 100 μM CDK1 inhibitor III or an equal volume of DMSO (refreshed daily) for 48 h. IE62 and cyclin B1 were detected by immunofluorescent confocal microscopy at the same laser and iris settings.

IE62 colocalizes with cyclin B1 in the Golgi apparatus of infected cells.

VZV spread was prevented by CDK1 inhibition, and so we investigated the localization of the kinase and its activator, cyclin B1, with VZV proteins. Previous studies indicated that cyclin B1 was cytoplasmic in VZV-infected HFFs, and the IE62 protein was both cytoplasmic and nuclear (28), and we therefore used confocal microscopy to observe their interactions with the Golgi apparatus. When IE62 was nuclear, early in infection, negligible cyclin B1 was detected (Fig. 2A, arrowheads). In addition, the morphology of the Golgi apparatus marker TGN46 was perinuclear and globular, its normal distribution in mock-infected cells (Fig. 2C) and uninfected cells surrounding the plaque. However, later in infection, when IE62 translocated to the cytoplasm, abundant cyclin B1 colocalized with IE62 and TGN46 (Fig. 2A and B, arrows). TGN46 morphology was aberrant in these late-infected cells, encircling nuclei, spreading throughout the cell, and extending in projections. Interestingly, cyclin B1 was not expressed in every infected cell and did not completely localize with IE62. A similar pattern was also observed in mock-infected HFFs, where cyclin B1 was diffuse in the cytoplasm of only scattered cells in the monolayer. Thus, cyclin B1 and IE62 colocalized primarily at later times during infection when IE62 accumulated in the cytoplasm, within or near the Golgi apparatus.

FIG. 2.

IE62 colocalizes with cyclin B1 in the cytoplasm of infected HFFs. Confluent HFF monolayers infected with VZV strain rPOka for 2 days (rows A and B) or mock-infected HFFs (row C) were processed for immunofluorescence confocal microscopy. Digital images show cyclin B1 (red), IE62 (green), TGN46 (blue) (trans-Golgi network), and the merge of all three channels (white) in a single, 2-μm plane. Different sources of antibodies were used to detect cyclin B1 and IE62 to rule out antigenic artifacts. Row A used MAb for cyclin B1 (sc-7393) and rabbit serum for IE62; row B used rabbit serum for cyclin B1 (sc-752) and MAb for IE62 (H6). An early-infected cell with nuclear IE62, negligible cyclin B1, and perinuclear TGN46 is visible (A, arrowheads) adjacent to a late-infected cell (IE62, cyclin B1, and TGN46 in the cytoplasm [A, arrows]). Areas of trilocalization appear white (B, arrows). Magnifications, ×400 (A) and ×1,000 (B and C).

IE62 is phosphorylated by CDK1/cyclin B1 in vitro.

The amino acid sequence of IE62 contains 24 putative CDK phosphorylation sites ([S/T]P) and one cyclin binding motif (RXL) in previously identified functional domains (Fig. 3C) (50). We hypothesized that IE62 served as a substrate for at least one CDK/cyclin complex and chose to examine CDK1/cyclin B1 due to the increased level of expression and associated kinase activity of cyclin B1 in infected cells, the colocalization of IE62 and cyclin B1, as well as the detrimental effects of CDK1 inhibition on VZV replication (28). IE62 phosphorylation by purified CDK1/cyclin B1 was investigated using in vitro kinase assays. IE62 fragments (Fig. 3C) spanning the entire amino acid sequence N-terminally fused with MBP were produced in protease-deficient E. coli cells and purified by affinity chromatography (10, 22). Fusion proteins were detected by immunoblotting in all of the kinase assay reaction mixtures to determine size and amount. However, they did not migrate through SDS-PAGE gels as predicted by length; larger fragments migrated to approximately the same size as smaller fragments. A likely explanation is that the prevalence of highly charged amino acids in certain domains affected the mobility of some peptide fragments. The IE62 expression plasmids were sequenced for integrity, and the use of protease inhibitors at all steps of the purification process ensured that little degradation was occurring. The in vitro kinase assays showed that MBP-IE62 fragments were differentially phosphorylated by purified, recombinant CDK1/cyclin B1 (Fig. 3A, top, asterisks). There was no phosphorylation of amino acids (aa) 823 to 1238 or the unconjugated MBP alone, which was expected due to the absence of consensus sequences for CDK phosphorylation. The C-terminal portion of IE62 (aa 1238 to 1310) was phosphorylated by CDK1/cyclin B1 despite the absence of consensus sequences; this IE62 domain includes a serine-rich acid patch that could serve as a nonspecific substrate for a serine/threonine kinase such as CDK1/cyclin B1. Fragments spanning the N-terminal portion of IE62 were phosphorylated by CDK1/cyclin B1, including aa 1 to 161, 162 to 505, and 505 to 823. Recombinant cyclin B1 was phosphorylated by CDK1 and yielded a band that migrated above the fusion proteins (Fig. 3A, arrow). These results identified IE62 domains that were preferentially phosphorylated by CDK1/cyclin B1 and prompted a more detailed evaluation of possible CDK target sites in IE62.

FIG. 3.

IE62 is phosphorylated by CDK1/cyclin B1 at multiple sites. (A) Kinase assays were performed using MBP fused to the N terminus of IE62 fragments as a substrate for recombinant active CDK1/cyclin B1; the addition of ³²P to the MBP-IE62 fusion peptides (asterisks) and recombinant cyclin B1(arrow) was detected using autoradiography (top). Immunoblots (IB) for MBP showed the presence of fusion protein in each reaction mixture (bottom) (asterisks indicate full-length peptide). (B) Mutated MBP-IE62 fusion proteins (immunoblot) (bottom) were tested in kinase assays (top) (asterisks, fragments; arrows, recombinant cyclin B1). Values given under the autoradiograph were calculated by normalizing phosphate incorporation to protein levels; 1.0 was set as the level of unmutated fragment phosphorylation, and asterisks indicate statistical significance (P < 0.05 by the Student t test). Kinase assays were repeated at least twice. MW, molecular weight (in thousands). (C) Diagram of the full-length IE62 protein showing mutated residues and certain IE62 functional areas (underlined). NLS, nuclear localization signal.

By analyzing the IE62 fragment protein sequence using PhosphoBase (8), possible CDK phosphorylation sites with high probability were identified as T10, S141, S245, and T680 and were mutated to alanine to block phosphorylation. Serines in the C-terminal portion of the protein were not mutated due to the lack of a consensus sequence and a high number of serines (18% of the amino acids). Assessments of these mutant peptides in kinase assays with recombinant CDK1/cyclin B1 suggested that two residues were the main targets for the kinase. Specifically, T10A resulted in a 60% reduction in phosphorylation, and S245A abrogated phosphorylation to background levels (Fig. 3B, top). These values were mathematically corrected for unequal protein loading (Fig. 3B, bottom) and were consistently reproduced over three separate protein purifications and kinase assays (data not shown). A double mutant consisting of T10A and S141A did not reduce phosphorylation below that of T10A alone, and the single mutation of S141A increased phosphorylation. Phosphorylation of the T680A fragment was reduced to 50% of that of the wild-type sequence, suggesting that it is also a target for the kinase in vitro, although the extent of phosphorylation decrease varied across experiments. This indicates that multiple portions of IE62 were phosphorylated by CDK1/cyclin B1 to various degrees, but S245 is the major phosphoacceptor site in the N-terminal portion of IE62, with T10A also being phosphorylated.

IE62 and cyclin B1 are present in a complex in infected HFFs.

We then assessed whether cyclin B1 and IE62 interact in VZV-infected cells by coimmunoprecipitation. Previous studies indicated that cyclin B1, typically expressed during G₂ phase, is highly induced in confluent HFFs (G₀/G₁ phase) infected with VZV (28). Here, this finding was repeated; cyclin B1 was detected by immunoblotting in a lysate of VZV-infected HFFs (Fig. 4, lane 1). Control immunoprecipitations were performed to exclude the nonspecific binding of IE62 and cyclin B1 to rabbit immunoglobulin G (IgG) (Fig. 4, lane 2), and conversely, lysates of HFFs arrested in G₂ phase yielded plentiful cyclin B1 with the specific rabbit antiserum (lane 3). When cyclin B1 was immunoprecipitated from infected cell lysates, IE62 was detected by the subsequent immunoblotting of the immunoprecipitates (Fig. 4, lanes 4 and 5). This interaction was also seen in the presence of ethidium bromide, which would disrupt DNA-protein interactions and release any proteins bridged by DNA (Fig. 4, lane 6) (27, 53). This strongly suggests that IE62 and cyclin B1 proteins are in a complex.

FIG. 4.

Cyclin B1 and IE62 form a complex in infected cells. Immunoblots (IB) for IE62 (top) and cyclin B1 (bottom) showed their presence in VZV-infected cell lysate or immunoprecipitates (IP). VZV-infected cell lysate (lane 1, 1/10 volume of the immunoprecipitation input), VZV-infected HFF immunoprecipitated with normal purified rabbit IgG (rIgG) (lane 2), G₂-synchronized HFFs (lane 3), or VZV-infected HFFs (lanes 4 to 6) immunoprecipitated with cyclin B1 without ethidium bromide (EtBr) (lane 5) or with 50 μg/ml ethidium bromide (lane 6).

CDK1/cyclin B1 phosphorylates full-length viral IE62.

To determine whether IE62 and cyclin B1 protein interactions could lead to phosphorylation in infected cells, we used full-length IE62 made in VZV-infected cells as a substrate for the kinase. IE62 is a substrate for the viral kinases ORF47 and ORF66 as well as cellular kinases (10, 20, 37). Indeed, substantial kinase activity was associated with IE62 following immunoprecipitation (Fig. 5, lane 1). To suppress this bound kinase activity, IE62 immunoprecipitates were heat inactivated for 10 min at 60°C prior to incubation with [³²P]ATP (13), which reduced IE62 phosphorylation by over 80% (Fig. 5, lane 2). This substrate was phosphorylated by recombinant active CDK1/cyclin B1 (Fig. 5, lane 3, arrow). A negative control immunoprecipitation with rabbit IgG confirmed the specificity of the immunoprecipitation (Fig. 5, lane 4). This demonstrates that native IE62 is a substrate for purified CDK1/cyclin B1 and reinforces the in vitro kinase assay results using IE62-MBP fusion proteins.

FIG. 5.

CDK1/cyclin B1 phosphorylates native IE62 protein. IE62 was immunoprecipitated (IP) from VZV-infected HFFs and used as a substrate in CDK1 assays. IE62 phosphorylation was detected by autoradiography for ³²P (arrow). Endogenous kinase activity associated with IE62 was immunoprecipitated from VZV-infected HFFs using rabbit serum against IE62 (lane 1), heat inactivation (H.I.) of kinase activity associated with IE62 immunoprecipitation (lane 2), heat inactivation followed by the addition of recombinant CDK1/cyclin B1 to IE62 immunoprecipitations (lane 3), and VZV-infected HFFs immunoprecipitated with rabbit IgG (rIgG) and then CDK1/cyclin B1 added to the kinase assay (lane 4).

CDK1 and cyclin B1 are present in virions.

IE62 is in the virion tegument (a proteinaceous layer between the outer envelope and inner nucleocapsid) (25). We hypothesized that CDK1 and/or cyclin B1 could also be present in virions, since they bind and phosphorylate IE62 and colocalize near the Golgi apparatus, where virion assembly occurs. Virions were purified from infected HFFs and analyzed for integrity, DNA content, and protein composition. The nature of VZV assembly within cultured cells differs from that of other alphaherpesviruses in that the vesicles containing virions are targeted to prelysosomes, and most are degraded prior to release from the cell (6, 12). The protocol for purifying virions involves separating these vesicles from other cellular components, followed by differential centrifugation over a Ficoll gradient; however, these virions are noninfectious (data not shown). To ensure that the preparation contained virions, negative-stain electron microscopy was performed. Mostly intact virions were observed (Fig. 6A). Particles consisted of fully enveloped virions with an electron-dense core and a surrounding membrane with a diameter of approximately 200 nm. Some smaller debris was seen, approximately 40 nm in diameter, which could be degraded virion elements that aggregated into long chains, as the frequency of their appearance increased with freeze-thaw cycles of the preparation (data not shown). There was no evidence of cellular vesicles or nuclei, reinforcing the purity of the preparation. PCR was used to indicate the presence of viral DNA and the absence of cellular DNA, and while IE62 DNA was amplified from infected HFF nuclear samples, virions, and the plasmid control, β-actin was amplified only from cellular DNA (Fig. 6B). Thus, the purified virus preparation contained VZV DNA, was not contaminated with measurable cellular DNA, and consisted of relatively pure virions.

FIG. 6.

CDK1 and cyclin B1 in virions. (A) Negative-stain electron microscopy of purified virions. Scale bar, 500 nm. (B) PCR analysis of purified virions, VZV-infected HFF nuclei (nuclei), a plasmid encoding IE62 (pCMV-62), and distilled water (dH₂O) using primers for IE62 and β-actin genes. The DNA ladder indicates fragment sizes (bp). (C) Protein lysates from mock-infected HFFs, VZV-infected HFFs, and purified virions were separated by SDS-PAGE and analyzed by immunoblots with the indicated antibodies. (D) Virions were treated with trypsin, detergents, or both to fractionate the virions. Proteins were separated by SDS-PAGE and immunoblotted for the indicated proteins, or the gel was stained with Coomassie for total protein content. S, supernatant; P, pellet; MCP, major capsid protein. (E) Purified virions were solubilized with detergent, and endogenous kinase assays were performed with [³²P]ATP. Reactions contained 50 μM CDK1 inhibitor III (CDK1 Inh.) or an equal volume of DMSO. The phosphorylation of IE62 was detected by autoradiography, and the level was determined by densitometry. The extent of phosphorylation in reactions with DMSO alone was set as 1.0 and compared to reactions with inhibitor to calculate the change.

The virion protein content was analyzed by immunoblotting in comparison to lysates of mock-infected, confluent HFFs as a negative control and VZV-infected HFFs as a positive control. It is known that the IE62, IE63, and ORF4 proteins are in virions, whereas ORF61 and ORF29 (single-stranded DNA-binding protein) are not (22, 25). As expected, virions contained the IE62, IE63, and ORF4 proteins and gE, the major envelope protein (Fig. 6C). In addition, immunoblots for ORF29 (the homologue of herpes simplex virus type 1 [HSV-1] ICP8) and ORF61 were consistently negative. Notably, immunoblots for p27 clearly showed that this cellular protein was absent from virions, although its level of expression increased upon VZV infection. p27 is a CDK binding protein that we have observed in the cytoplasm of VZV-infected HFFs (28). Its absence from virions was taken as an indication that cytosolic proteins were not adhering to or being assembled into virions in a nonspecific manner. Strikingly, virions also contained CDK1 and cyclin B1 (Fig. 6C). This finding was verified using four separate preparations of purified virions, two from infected HFFs and two from infected MeWo cells, a transformed cell type permissive for VZV (data not shown).

CDK1 and cyclin B1 are in the virion tegument.

We next determined where in the virions these cellular proteins were located. Using trypsin to digest proteins into small peptides and detergents to permeabilize membranes, the virions were fractionated into discrete components. Trypsin treatment alone determined whether the cellular proteins adhered to the exterior of the virion or if they were within the viral envelope. gE was used as a positive control, as it is in the viral envelope; an antibody against the ectodomain of gE showed by immunoblotting that it was digested by trypsin (Fig. 6D). IE62, IE63, and ORF4 remained intact following trypsin treatment, confirming their tegument location. Similarly, CDK1 and cyclin B1 were also protected from trypsin, suggesting localization in the tegument and indicating that they were not stuck to the exterior of the virion.

Nonionic Triton X-100 and anionic sodium deoxycholate detergents were used to permeabilize the viral envelope and evaluate protein solubility. No proteins were soluble with a mock detergent treatment, demonstrating a lack of freely diffusible proteins (Fig. 6D). Detergent released gE from the envelope membrane, and it became completely soluble. IE62 and IE63 became partially soluble, mimicking the actions of VP16 (the HSV-1 homologue of ORF10), a loosely tethered tegument protein. ORF4 remained in the insoluble fraction following detergent treatment. Interestingly, CDK1 and cyclin B1 became completely soluble following detergent treatment. Thus, CDK1 and cyclin B1 are inside the envelope, since they are resistant to trypsin but are sensitive to detergents.

Finally, detergent and trypsin treatments were combined to digest both envelope and tegument proteins. This treatment permeabilizes the viral membrane, allowing trypsin to digest all proteins between the envelope and nucleocapsid but not the tightly constructed viral nucleocapsid itself. Following treatment with detergent plus trypsin, gE, IE62, IE63, ORF4, and also CDK1 and cyclin B1 were digested. However, not all proteins were digested with the dual treatment; a 150-kDa protein, presumably the viral major capsid protein, remained in abundance following detergent solubilization and trypsin treatment as shown by Coomassie staining of the gel (Fig. 6D).

VZV-associated CDK1/cyclin B1 is kinetically active.

The surprising and unprecedented finding of CDK1/cyclin B1 in VZV virions raised the question of whether the kinase was active upon release from the particle. This was complicated by the presence of viral kinases in the virion (37) as well as other cellular kinases and the lack of infectivity of the purified virions (59). To evaluate the contribution of CDK1 only, CDK1 inhibitor III was used again for virion kinase assays. Virions were permeabilized with detergents, and kinase reactions were performed without the addition of exogenous histone H1 substrate to evaluate the phosphorylation of IE62 within the virion. Compared to kinase reactions with the drug diluent DMSO, IE62 phosphorylation decreased by 60% in the presence of CDK1 inhibitor III (Fig. 6E). The identity of this phosphoprotein was confirmed as being IE62 by immunoblotting of the membrane after the ³²P signal had decayed. The residual phosphorylation could be from ORF47 or ORF66 as well as other cellular kinases that were not affected by this inhibitor (59). Overall, these results demonstrate that active CDK1/cyclin B1 is located in the VZV tegument and phosphorylates IE62 under appropriate conditions.

Virion components cofractionate through a Ficoll gradient.

The incorporation of CDK1/cyclin B1 into tegument raised the question of how this occurred. It is possible that soluble proteins such as CDK1/cyclin B1 were engulfed during secondary envelopment of VZV capsids and tegument; in other words, VZV packaging merely sampled the cytosolic milieu, or, as we hypothesized, CDK1/cyclin B1 was selectively incorporated into virions by its association with IE62, an abundant tegument protein. A density fractionation approach was taken to determine whether CDK1/cyclin B1 was redistributed from the cytosol to membrane-bound compartments where secondary envelopment occurs, which is the route followed by IE62 and IE63 during tegument formation. Lysates from VZV-infected HFFs were prepared in the same way as was done for virion purification. In parallel, HFFs semisynchronized in G₂ phase (when cyclin B1 is most abundant) were lysed. The lysates were then layered over a 5 to 15% Ficoll gradient. Centrifugation separated the soluble proteins, which remained in the top fractions, from particles of increasing density, including membrane fragments, vesicles, virions, and debris. Two major bands were observed in the gradient of the VZV-infected lysates but not in control lysates, one located approximately 2 cm from the top of the tube and another located at 10 to 11 cm. Thirteen 1-ml fractions were drawn off the top of the gradient, and the pelleted debris was resuspended in the remaining liquid to assess the degraded virions that predominate in cultured cells. Immunoblots for cyclin B1, CDK1, p27, IE62, IE63, and gE were then performed to analyze protein distribution. In G₂-synchronized HFFs, cyclin B1 was most abundant in the top two fractions of the gradient, where the undiluted lysate was layered, and much less protein entered the Ficoll in fractions 3 to 6 (Fig. 7). Cyclin B1 was not detected deeper in the gradient or the pellet, indicating that it was in a low-density compartment. In contrast, cyclin B1 was scant in the top two fractions from lysates of VZV-infected HFFs, and it was redistributed to fractions 10 to 13 and the pellet. This distribution pattern was also observed for CDK1, IE63, and IE62, in agreement with previous results showing cyclin B1 associated with IE62. Unlike tegument proteins, gE was detected in all gradient fractions and the pellet, with protein concentrating at both high and low densities. This is due to its dispersal on vesicles and virions of all sizes, including endoplasmic reticulum, Golgi apparatus, trans-Golgi network, endosomes, multivesicular bodies, and plasma membrane. More extensive controls were performed to validate the separation of nuclei from the cytosolic lysates that were separated in the gradient (data not shown). As expected, the VZV single-stranded DNA-binding protein that localizes exclusively to the nucleus, ORF29, was not detectable in any fraction of the Ficoll gradient. The host cell protein p27, localized to the cytoplasm but not in virions (Fig. 6C), was detected in the top two fractions of the gradient and faintly in the pellet but not in association with the virion proteins IE62, IE63, or cyclin B1 in fractions 10 to 11. The finding that cyclin B1 cofractionated with tegument proteins and not with membrane glycoprotein suggests that it was selectively assembled into virions.

FIG. 7.

CDK1 and cyclin cofractionate with viral tegument proteins. Cytosol from G₂-semisynchronized or VZV-infected HFFs was fractionated on a Ficoll gradient from 5% (top) to 15% (bottom). Thirteen 1-ml fractions and the pellet (P) were taken sequentially from the top, and the proteins were then separated by SDS-PAGE and immunoblotted for the indicated proteins.

DISCUSSION

Our results demonstrate that VZV IE62 is a substrate for CDK1/cyclin B1. IE62 interacted with CDK1/cyclin B1 in vitro and in infected HFFs, which was not dependent on binding DNA. The interaction with IE62, and possibly other VZV proteins, led to the incorporation of active CDK1/cyclin B1 in the virion. The phosphorylation of IE62 peptides on residues T10, S245, and T680 validated the sites predicted by PhosphoBase analysis. The T10A and T680A mutations reduced phosphorylation by CDK1/cyclin B1; S245A completely abrogated phosphorylation. The amino acid sequence SPKK at positions 245 to 248 has been established as the optimal motif for both cyclin binding and phosphorylation by CDKs 1, 2, 4, and 6 (52). For cyclin binding at suboptimal CDK phosphorylation sites such as TPQR (aa 10 to 13) and TPRK (aa 680 to 683), an RXL sequence (amino acids RLL [positions 854 to 856]), called a Cy motif, is needed C terminal to the consensus [S/T]PXX site. Thus, S245 may be an exclusive target of CDK1/cyclin B1, while T10 and T680 are recognized by other kinases as well.

IE62 is an essential protein expressed from ORF62 and ORF71, and it is required for VZV replication in all cell types and tissues tested (46). The deletion of both copies was lethal, and the restoration of a single copy of ORF62 at a heterologous site in the genome resulted in a small-plaque phenotype. When this mutant strain was modified at S245A, no further reduction in plaque size was seen in MeWo cells. This point mutation, and other alterations in the genes for IE62, frequently reverted to the wild-type sequence (46). However, it was not determined whether the S245A mutation altered IE62 localization or binding partners, and due to the small-plaque phenotype and frequent reversal mutations, we could not use this mutant strain for virion purifications. This mutant virus was propagated in transformed MeWo cells, which express high levels of cell cycle kinases. These kinases could target other phosphorylation sites and compensate for the loss of S245. Many kinases, both viral and cellular, may target S245, and the affinity of CDK1/cyclin B1 for IE62 demonstrates a role in VZV biology.

Using a specific inhibitor of CDK1, we showed that a loss of kinase activity results in decreased viral replication. Although virus yield was not totally abolished (1-log decrease), other kinases such as CDK2 or JNK or the viral kinases ORF47 and ORF66 may have compensated for the loss of one kinase (28, 59). Interestingly, levels of cyclin B1 production appeared to be lower in cells treated with this inhibitor (Fig. 1B). This could be due to a decrease in the level of phosphorylation of cyclin B1 by CDK1, leading to its degradation (4, 19). Thus, CDK1 inhibition could modulate cyclin B1 protein, thus influencing viral protein phosphorylation with the effect of blocking viral replication.

In addition to the CDK1 phosphorylation of IE62, CDKs are likely involved at multiple stages of VZV replication. In previous studies, we found that the transcription of ORF62/ORF71 was blocked very early when cells were treated with roscovitine, a broad CDK inhibitor (51). Purvalanol, a similar CDK inhibitor, also prevented VZV growth (34). Possible explanations for the VZV dependence on CDK1 may be drawn from its known functions, which are nuclear lamina breakdown, DNA origin repression, and Golgi apparatus breakdown (9, 29, 30). We are currently investigating how increased CDK1 activity could result in Golgi apparatus morphology changes that occur following VZV infection, as seen in Fig. 2.

The presence of active CDK1/cyclin B1 in a virion is novel although not entirely unexpected based on preliminary work. By confocal immunofluorescence microscopy using the TGN46 marker, we identified cyclin B1 and IE62 colocalization in the trans-Golgi network of VZV-infected HFFs (Fig. 2). Since herpesvirus envelopes are derived from the trans-Golgi apparatus (12, 16), we speculate that cyclin B1 is incorporated into the virion via its subcellular localization and possible interactions with the Golgi membrane. Indeed, the cyclin B1 and CDK1 proteins were located in the tegument, since both proteins resisted trypsin treatment of purified virions. The treatment of virions with Triton X-100, which solubilizes proteins in membranes, released gE and CDK1/cyclin B1, whereas other tegument proteins were less soluble (Fig. 6D). A possible link between these three proteins is through gI, which forms a heterodimer with gE and is known to be phosphorylated by CDK1 (58). It will be interesting to investigate whether cyclin B1 tethers gE/gI and IE62 together during assembly (Fig. 8).

FIG. 8.

Proposed model for incorporation of CDK1 and cyclin B1 into the VZV tegument. Possible protein-protein interactions that could result in the incorporation of cellular CDK1 and cyclin B1 in the tegument are shown. gI forms a heterodimer with gE and has an identified CDK phosphorylation site in the endodomain. Cyclin B1 binds CDK1, and this study shows a direct interaction with IE62 protein. IE62 is a very large protein (∼175 kDa) that dimerizes and binds a plethora of viral proteins such as IE63 and IE4. The connections between these proteins and the capsid are unknown.

Viral protein phosphorylation by cellular kinases has been documented previously for herpesviruses. ICP4, the HSV-1 homologue of IE62, contains several sites that are phosphorylated both in vitro and in vivo (41, 55); protein kinase A could be responsible (56). These phosphorylation events are thought to influence ICP4 subcellular localization and transactivator function. Our data do not address whether the CDK1 phosphorylation of IE62 regulates location because the identified phosphorylation sites S245 and T10 are not in or near the nuclear localization sequence, but we do hypothesize that IE62 phosphorylation by CDK1/cyclin B1 could impact viral assembly. It was not feasible to study this due to technical barriers: the use of small interfering RNA directed against cyclin B1 or the CDK1 inhibitor decreased viral replication, so virions could not be isolated, and CDK1 or cyclin B1 knockout human cell lines are not available (S. Taylor and J. F. Moffat, unpublished observations).

It has long been understood that viruses depend on host functions for replication inside cells, and it is now becoming clear that cellular proteins may have undiscovered roles within the virus particle. Viruses manipulate many cellular proteins, such as CDKs, cyclins, mitogen-activated protein kinases, and DNA replication machinery, that are necessary for viral propagation, and the virus might benefit if these components were available immediately upon infection (17, 28, 42, 48). Incorporating active kinases and other host proteins into the virion could be advantageous for several reasons. First, membrane fusion and release of the tegument constitute the earliest stage when cell proteins in the virion could act. For instance, phosphorylation by host kinases could promote tegument dissolution and stimulate the migration of viral and cellular transcription factors to the nucleus (36). Second, host proteins brought in with the virus could make the intracellular environment more hospitable for viral entry or replication by affecting cell pathways for survival, the interferon response, or cytokine release. Third, cellular kinases in virions could allow VZV to replicate in a variety of cell types that are normally quiescent, including memory T cells and differentiated neurons. However, an understanding of how and why cellular proteins are brought into virions remains elusive, and few details are known. The technical aspect of purifying virions from intracellular compartments complicates the studies of the VZV proteome. Our studies have controlled for this by careful gradient fractionation as well as PCR and electron microscopy investigations to ensure the removal of measurable cellular contaminants. However, it cannot be ruled out that virion preparations are entirely free of cellular vesicles that are the same size and density as virions, and therefore, immunoelectron microscopy studies are being undertaken to show the specific incorporation of CDK1 and cyclin B1 into virions.

The presence of cellular proteins and kinases in purified virions is not a novel topic (reviewed in reference 32). Human immunodeficiency virus has been the most extensively studied virus with regard to cellular protein incorporation: virions contain extracellular signal-regulated kinase 2 (5), actin (40), and numerous heat shock proteins (14). The cellular proteome of HCMV has also been investigated. Polo-like kinase 1 is incorporated into the virion via an interaction with the viral protein pp65 (11), and cellular microtubule elements have been found in the form of an actin-related protein (1). In addition, CKII is also packaged into HCMV virions, with a hypothesized role in the phosphorylation of IκBα to induce the activation of NF-κB for viral immediate-early protein expression (39). The related animal alphaherpesvirus PRV virion contains cellular proteins, including β-actin, and there is a possible role for nuclear actin in PRV capsid movement in the nucleus (L. Enquist, unpublished data; 54). Purified PRV virions also contain HSP70 and annexin A2 for as-yet-unidentified functions (33). This emerging field of cell proteins as structural components of viruses will hopefully yield information about virion assembly and protein-protein interactions as well as new targets for antiviral drugs.