Antagonistic Relationship between Human Cytomegalovirus pUL27 and pUL97 Activities during Infection

ABSTRACT

Human cytomegalovirus (HCMV) is a member of the betaherpesvirus family. During infection, an array of viral proteins manipulates the host cell cycle. We have previously shown that expression of HCMV pUL27 results in increased levels of the cyclin-dependent kinase (CDK) inhibitor p21^Cip1. In addition, pUL27 is necessary for the full antiviral activity of the pUL97 kinase inhibitor maribavir (MBV). The purpose of this study was to define the relationship between pUL27 and pUL97 and its role in MBV antiviral activity. We observed that expression of wild-type but not kinase-inactive pUL97 disrupted pUL27-dependent induction of p21^Cip1. Furthermore, pUL97 associated with and promoted the phosphorylation of pUL27. During infection, inhibition of the kinase resulted in elevated levels of p21^Cip1 in wild-type virus but not a pUL27-deficient virus. We manipulated the p21^Cip1 levels to evaluate the functional consequence to MBV. Overexpression of p21^Cip1 restored MBV activity against a pUL27-deficient virus, while disruption reduced activity against wild-type virus. We provide evidence that the functional target of p21^Cip1 in the context of MBV activity is CDK1. One CDK-like activity of pUL97 is to phosphorylate nuclear lamin A/C, resulting in altered nuclear morphology and increased viral egress. In the presence of MBV, we observed that infection using a pUL27-deficient virus still altered the nuclear morphology. This was prevented by the addition of a CDK inhibitor. Overall, our results demonstrate an antagonistic relationship between pUL27 and pUL97 activities centering on p21^Cip1 and support the idea that CDKs can complement some activities of pUL97.

IMPORTANCE HCMV infection results in severe disease upon immunosuppression and is a leading cause of congenital birth defects. Effective antiviral compounds exist, yet they exhibit high levels of toxicity, are not approved for use during pregnancy, and can result in antiviral resistance. Our studies have uncovered new information regarding the antiviral efficacy of the HCMV pUL97 kinase inhibitor MBV as it relates to the complex interplay between pUL97 and a second HCMV protein, pUL27. We demonstrate that pUL97 functions antagonistically against pUL27 by phosphorylation-dependent inactivation of pUL27-mediated induction of p21^Cip1. In contrast, we provide evidence that p21^Cip1 functions to antagonize overlapping activities between pUL97 and cellular CDKs. In addition, these studies further support the notion that CDK inhibitors or p21^Cip1 activators might be useful in combination with MBV to effectively inhibit HCMV infections.

INTRODUCTION

Human cytomegalovirus (HCMV) infects the majority of the world's population (1). Infection of immunocompetent children and adults is usually asymptomatic or associated with minor disease. In contrast, HCMV infection in immunocompromised patients results in serious disease, especially in organ transplant recipients receiving immunosuppressants (2). HCMV is also the leading congenital infection in the developed world (3). Currently, the approved antiviral pharmaceuticals manage infection well, though toxicity and bioavailability remain concerns for their clinical application (2). However, HCMV can rapidly develop resistance to antiviral treatment through selected genetic mutations (4). Understanding the mechanisms of resistance to the available drugs is crucial to identifying therapy regimens that surmount resistance.

The HCMV serine/threonine kinase pUL97 is a kinase that is conserved among the members of the herpesvirus family. The kinase is expressed with early late kinetics and is incorporated into the tegument (5, 6). pUL97 is not essential for viral replication, but a loss of kinase activity through genetic or pharmaceutical means results in severe attenuation of replication (7, 8). The kinase has multiple functions during infection that are important for viral replication (reviewed in reference 9). It has been shown to function in promoting viral gene expression, stimulating viral DNA (vDNA) synthesis, nuclear egress of the viral nucleocapsid, and formation of the cytoplasmic assembly compartment (9). pUL97 targets multiple viral and cellular proteins for phosphorylation, including overlapping targets with cellular cyclin-dependent kinases (CDKs) (10–12). For these reasons, pUL97 has been designated a viral CDK-like kinase (13). CDK-like activities include phosphorylation of pRB, possibly to stimulate cell cycle regulatory pathways important for viral replication (11, 14–16), and phosphorylation of A- and C-type lamins, which induces nuclear lamina disassembly and facilitates nucleocapsid egress (8, 10, 12).

pUL97 is an important enzymatic target for pharmaceutical antiviral therapeutics due to its numerous roles during infection. Maribavir (MBV) is a selective pUL97 inhibitor that demonstrates high oral bioavailability and low toxicity (17–20). It has undergone several clinical trials, been given orphan drug status, and could be useful for treating infections refractory to other antivirals (21). Passage of virus in cell culture in the presence of MBV selects for resistant mutants (reviewed in reference 22). Mutations that confer resistance have been mapped to the UL97 locus as well as UL27 (22–27). Interestingly, mutations in UL97 that disrupt kinase activity also promote mutations in UL27 (22). pUL27 remains largely uncharacterized. Expression occurs in the nucleus with nucleolar localization (28, 29), and MBV-associated mutations in UL27 result in altered localization (29). Our lab has previously demonstrated that pUL27 functions to increase the levels of the CDK inhibitor protein p21^Cip1 and arrest cells in G₀/G₁ (28). This is mediated in part by the pUL27-dependent degradation of Tip60, an acetyltransferase (28, 30). In silico modeling of pUL27 suggests that it might interact with and be a target of pUL97 (31), but the relationship between these two viral proteins remains unknown.

HCMV manipulates cell cycle regulatory pathways during infection (reviewed in reference 32). Cells in G₀ or G₁ are permissive to initiation of viral immediate early gene expression during lytic infection. In contrast, expression is inhibited during S, G₂, and M and involves an interaction between cyclin A and the viral protein pp150 (33, 34). Lytic infection with HCMV results in a pseudo-cell cycle phase characterized by inhibition of cellular DNA synthesis along with processes that normally occur during the G₁/S transition or mitosis (35–42). For example, during infection pRB is inactivated by phosphorylation, which allows E2F-mediated induction of target genes, such as DNA synthesis factors (16, 35, 43). CDK1 and CDK2 steady-state levels increase along with steady-state cyclin B and E levels and associated kinase activity (35, 44). During infection, Myt1 and Wee1 are degraded, which relieves inhibition of the Cdc25 phosphatase and CDK activity (44). The CDK inhibitor p21^Cip1 is also tightly regulated, with steady-state levels increasing and being maintained early during infection, followed by degradation at later times (45–47). The decrease in p21^Cip1 is mediated in part by proteasome- and calpain-dependent protein degradation (47) as well as decreased RNA expression (47–49).

In this study, we demonstrate that pUL97 antagonizes pUL27-mediated induction of p21^Cip1. Furthermore, increased p21^Cip1 expression or altered CDK activity restores MBV antiviral activity in the absence of pUL27. Taken together, our results identify an antagonistic relationship between pUL27 and pUL97 and provide insight into pathways important for MBV antiviral activity.

MATERIALS AND METHODS

Biological reagents.

The wild-type HCMV strain AD169 (ADwt) was obtained from the AD169 bacterial artificial chromosome (BAC) clone (50). A UL27 substitution mutant, ADdel27 (ADd27) was constructed using BAC recombineering as previously described (28). ADd27 and matched AD169 stocks were prepared by cotransfecting BAC DNA with the pCGN-pUL82-HA expression vector. Viral stocks were concentrated by collecting culture medium and cell lysate and pelleting through a sorbitol cushion (20% d-sorbitol, 50 mM Tris-HCl, pH 7.2, 1 mM MgCl₂) at 55,000 × g for 1 h using a Sorvall WX-90 ultracentrifuge and SureSpin 630 36 ml rotor (Thermo Scientific). Viral stock titers were obtained by infecting human foreskin fibroblasts (HFFs) with eight 10-fold serial dilutions of the viral stock and counting the number of IE1-positive cells at 48 h postinfection (hpi). The numbers of IE1-positive cells per dilution series were counted and defined as the number of infectious units (IU) per milliliter. To estimate viral entry, HFFs were infected at 1 IU/cell for 2 h. Cells were fixed and permeabilized by incubation with 100% methanol at −20°C for 30 min. Fixed cells were incubated for 1.5 h with antibody against pp65 (1:500 dilution in 3% bovine serum albumin [BSA]–phosphate-buffered saline [PBS] containing 0.1% Tween 20 [PBS-T]), washed with PBS-T, and then incubated with anti-mouse IgG Alexa Fluor 488 (1:1,000 dilution in 3% BSA–PBS-T) for 1 h. Fifty to 100 cells were quantified for nuclear pp65 using the 20× objective lens on a Nikon Eclipse TS100 inverted microscope (Nikon Inc., Melville, NY).

HFFs, U-2 OS osteosarcoma (U2OS) cells, and U373 astrocytoma cells were propagated in Dulbecco's modified Eagle medium (DMEM) (Life Technologies) supplemented with 7% fetal bovine serum (FBS; non-United States-qualified FBS; Life Technologies) and 1% penicillin-streptomycin (Life Technologies). All transfections and infections were performed following 24 h of serum starvation (0.1% or 0.5% serum) of cells grown to ∼75% confluence for HFFs and U373 cells. Confluence was estimated using a hemocytometer.

Cells were infected at a multiplicity of infection (MOI) of 1 IU/cell in DMEM (with 0.1% or 7% FBS). Drug was applied at the start of infection. Cells were washed at 2 hpi with Dulbecco's PBS, and new medium was reapplied. Compounds were added to the culture medium at the start of infection and reapplied after the inoculum was removed. Cells were treated with 10 μM maribavir (provided by ViroPharma, now Shire plc), 5 μM roscovitine (Calbiochem), 120 and 300 nM CDK2 inhibitor II (Santa Cruz), 75 and 150 nM 2-cyanoethyl alsterpaullone (alsterpaullone; Santa Cruz), 2.5 μM CDC25 phosphatase inhibitor II (Santa Cruz), or dimethyl sulfoxide (DMSO) as a vehicle (Sigma). Electroporation of plasmid or viral BAC DNA was carried out in DMEM at 260 mV for 23 ms in a 4-mm electroporation cuvette. Cells were transfected using the Fugene 6 reagent (Promega) following the manufacturer's protocol. The pCGN-UL82-HA, pCGN-UL97-HA, and pCGN-UL97K355M-HA plasmids were generously provided by Robert Kalejta (University of Wisconsin, Madison) (13). The pCGN-UL27-HA and pLXSN-UL27-myc plasmids were developed as previously described (28). pEGFP-UL27, pEGFP-UL27R233S, and pEGFP-UL27aa1-415 were generously provided by Morgan Hakki and Sunwen Chou (Oregon Health Sciences University) (29). The pLL3.7-CDK1AF-FLAG and pLL3.7 vectors were generously provided by Liu Yang (University of Washington) (51). Disruption of p21^Cip1 expression was performed by transfecting cells that were serum starved for 24 h with p21^Cip1 small interfering RNA (siRNA; Cell Signaling) or with scrambled siRNA (Cell Signaling) using the Fugene 6 reagent according to the manufacturer's recommendations. Assays with vector controls were performed using a pCGN, pLXSN, or pLL3.7 vector.

The following antibodies were used for Western blot (WB), immunoprecipitation (IP), or immunofluorescence (IF) analysis: mouse anti-FLAG M2 (WB and IP analyses; Sigma), mouse anti-GAPDH (anti-glyceraldehyde-3-phosphate dehydrogenase; clone 0411; WB analysis; Santa Cruz Biotechnology), mouse antihemagglutinin (anti-HA; clone HA-7; WB and IP analyses; Sigma-Aldrich), rabbit anti-Myc (WB and IP analyses; Cell Signaling), mouse anti-green fluorescent protein (anti-GFP; WB and IP analyses; Santa Cruz), mouse anti-p21^Cip1 (WB analysis; Millipore), mouse anti-lamin A/C (IF analysis; Millipore), antiphosphoserine (WB analysis; Millipore), and antiphosphothreonine (WB analysis) and mouse control IgG (Santa Cruz Biotechnology). The antibodies against HCMV proteins were mouse anti-pUL123 (WB and IF analyses), mouse anti-pUL122 (WB analysis), and mouse anti-pUL83 (WB analysis), which were generously provided by Tom Shenk (Princeton University). Mouse anti-pUL97 (WB analysis) was kindly supplied by Mark Prichard (University of Alabama at Birmingham). Mouse anti-pUL44 (WB analysis) was purchased from Virusys. Secondary antibodies for Western blot analysis utilized goat anti-mouse immunoglobulin conjugated with horseradish peroxidase (HRP) and donkey anti-rabbit immunoglobulin conjugated with HRP (Jackson ImmunoResearch), and the secondary antibody for immunofluorescence was donkey anti-mouse IgG (H+L) conjugated with Alexa Fluor 488 (Life Technologies).

Analysis of protein and DNA.

Western blot analyses of steady-state protein levels were performed on cells lysed by sonication in a lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% NP-40). Protein from lysates was resolved by sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis (SDS-PAGE) and transferred by semidry transfer to a Protran nitrocellulose membrane (Whatman Inc.). Blocking was completed in 5% milk or 3% BSA in PBS-T. The blocked membrane was incubated in primary antibody diluted in 5% milk or 3% BSA–PBS-T, followed by a secondary antibody conjugated to HRP diluted in 5% milk in PBS-T. Antibodies were detected with enhanced chemiluminescence (ECL; GE Healthcare) and film. Immunoprecipitation experiments were initiated by lysing cells using sonication in a lysis buffer. Immunoprecipitation was performed using protein G Dynabeads (Life Technologies). Lysates were precleared for 30 min at 4°C with the beads. The beads were washed twice in lysis buffer, bound to antibody for 30 min at room temperature, and then washed twice with lysis buffer. Samples were incubated with antibody for 3 h at 4°C with rotation. The beads were washed 3 to 5 times with lysis buffer and resuspended in Laemmli sample buffer. The immunoprecipitated samples and 5% of the lysate were assessed by Western blotting. All buffers used for lysis or washes contained cOmplete, mini, EDTA-free protease inhibitor cocktail (Roche) and PhosStop (Roche).

The vDNA content from infected cells was evaluated using quantitative real-time PCR as previously described (52). The primers used to quantify the vDNA and cellular DNA contents included UL123 (5′-GCCTTCCCTAAGACCACCAAT-3′ and 5′-ATTTTCTGGGCATAAGCCATAATC-3′) and GAPDH (5′-ACCCACTCCTCCACCTTTGAC-3′ and 5′-CTGTTGCTGTAGCCAAATTCGT-3′), respectively. DNA quantification was performed using a 7900HT Fast real-time PCR system (Applied Biosystems). AD169 wild-type-infected cells, collected at the time point being analyzed, were used as a sample to determine a standard curve for each primer set. The relative quantities of the samples were determined from the standard curve. Viral DNA was normalized to the relative quantities of GAPDH for each sample.

Cell cycle analysis.

Cell cycle analysis was performed as previously described (28). Subconfluent U2OS cells were serum starved with 0.5% serum for 24 h and were then transfected with scrambled or p21^Cip1-specific siRNA as described above. At 24 h after transfection with siRNA, the cells were then transfected with the pCGN-UL27-HA vector or the pCGN empty vector. At 24 h, the cells were released from starvation, placed into medium with 7% serum containing 75 ng/ml of nocodazole, and allowed to grow for 16 h. The cells were harvested and fixed in 70% ethanol. The fixed cells were permeabilized with 0.1% Triton X-100 for 5 min and then probed with anti-HA antibody for 1 h at 25°C. The cells were washed twice with PBS-T, and then secondary antibody conjugated to Alexa Fluor 488 was added to the fixed cells for 30 min at 25°C and the cells were washed twice in PBS-T. The cells were treated with cell cycle reagent (Millipore) in the dark for 30 min. Analysis of the DNA content was performed on a Guava EasyCyte mini-flow cytometer (Millipore). Cells expressing the viral protein were distinguished from cells not expressing the protein by gating for viral protein expression. The flow cytometry data were analyzed for cell cycle using FlowJo analysis software.

MS.

For transfection experiments, approximately 2 × 10⁸ U2OS cells were lysed for immunoprecipitation. For infection, approximately 3 × 10⁸ HFFs were lysed for immunoprecipitation. Immunoprecipitation was performed as described above with the following modification: the beads were washed 3 times with lysis buffer, 1 time with 1:1 lysis buffer-PBS, and then 2 times with PBS. Elution was performed by incubating the beads in 0.4 M NH₄OH for 20 min at 25°C with agitation. Disulfide bonds were reduced in 10 mM dithiothreitol at 25°C for 30 min. Free cysteines were alkylated in 10 mM iodoacetamide at 25°C for 30 min. In-solution digestion was performed with 1 μg of trypsin gold, mass spectrometry (MS) grade (Promega), at 37°C for 12 to 15 h. Digestions were stopped by trifluoroacetic acid (TFA) added to a final concentration of 0.1%. Peptides were bound to Poros Oligo R3 bulk medium (Applied Biosystems), which was loaded into narrow-bore GELoader tips (Eppendorf). The columns were washed with 0.1% TFA and eluted using 0.1% TFA–70% acetonitrile.

Peptides that had been cleaned with Oligo R3 were enriched for phosphopeptides using Titansphere TiO₂ beads (GL Sciences). TiO₂ beads were incubated with the peptides at 21°C for 15 min with gentle agitation. The supernatant was removed, and the incubation with the TiO₂ beads was repeated. The TiO₂ beads from both incubations were combined and washed once with 1% TFA–80% acetonitrile and then once with 0.1% TFA–20% acetonitrile. Bound peptides were eluted with 1% NH₄OH for 15 min at 21°C with vigorous agitation. The eluate was then cleaned by repeating the cleaning with Oligo R3 as described above.

Mass spectral analysis was performed on an LTQ-Orbitrap Velos mass spectrometer (Thermo Scientific) as described by Bigley et al. (19). Briefly, full-scan MS spectra (m/z 300 to 2,000) were recorded in the Orbitrap analyzer. The resolution was set at 30,000 for MS/MS of the 10 most intense peptide ions in the linear ion trap analyzer. Neutral loss-triggered multistage activation for simultaneous fragmentation of the neutral loss product and precursor was enabled at m/z −98, −49, and −32.6 relative to the precursor ion, corresponding to a neutral loss of the phosphate moiety from +1, +2, and +3 charged ions, respectively. The spectra were analyzed using MaxQuant software (version 1.2.2.5) (53) and the UniProtKB Homo sapiens reference proteome database, containing 70,136 canonical and isoform sequences (retrieved in February 2013). HCMV reference proteomes for strains TB40/E, FIX, AD169, and Toledo were added and used for HCMV peptide identification. MaxQuant default parameters were used where applicable. Variable modifications included phosphorylation, protein N-terminus acetylation, and methionine. The digestion mode was set as full trypsin specificity, with missed cleavages being set to 2. The minimum peptide consideration was set to 7 amino acids, and the peptide and protein false discovery rate was set to 1%. The product ion mass tolerance was set to 20 ppm. All reported sites of phosphorylation were manually verified, with the spectrum for each site being presented in Fig. 3 and the supplemental material.

FIG 3.

HCMV pUL97 expression influences pUL27 phosphorylation. (A) Expression vectors for wild-type pUL27-HA or pUL97-HA or an empty vector were transfected into U2OS cells. Samples were immunoprecipitated using an HA-specific antibody and evaluated by Western blotting using the indicated antibodies or on gel using Pro-Q Diamond staining and Coomassie staining. The numbers to the left of the gels are molecular masses (in kilodaltons). (B) HFFs were grown in 0.5% serum for 24 h, released into 7% serum, mock infected or infected with ADwt or AD27F virus at an MOI of 1 IU/cell, and collected at the indicated times. Samples were immunoprecipitated using a FLAG-specific antibody and evaluated by Western blotting using the indicated antibodies or on gel using Pro-Q Diamond staining and Coomassie staining. (C) pUL27-HA was immunoprecipitated from U2OS cells transfected with pUL27-HA and pUL97-HA. Eluents were digested with trypsin and enriched using TiO₂, and phosphopeptides were identified using an LTQ-Orbitrap mass spectrometer. The phosphorylated amino acid is noted (pT or pS in the peptide sequence indicates the site of phosphorylation), and the spectra are presented in the supplemental material. The pUL97 consensus sites are indicated (*). Spectra confirming the phosphorylation of pUL27 at threonine 78 during infection are shown. Circles indicate sites of phosphorylation, and diamonds identify a subset of previously observed MBV-selected mutations with C415st indicating the stop mutant. (D) Phosphorylation on peptides from pUL97, including serine 185, was also observed. Amino acid numbers separated by commas indicate that phosphorylation may occur on either site. The observed y and b ions are listed. Additional spectra are presented in the supplemental material.

Immunofluorescence for nuclear morphology.

Cells were plated at ∼50% confluence in 6-well dishes containing glass coverslips. Immunofluorescence analysis was initiated by fixing the cells with 4% paraformaldehyde at room temperature (RT) for 20 min, and the cells were then permeabilized in 0.1% Triton X-100 at RT for 30 min and blocked with 3% BSA–PBS-T for 30 min. Primary antibody was diluted in 3% BSA–PBS-T, the mixture was added to the cells for 1.5 h at RT, and the cells were washed 3 times with PBS-T. Secondary antibody conjugated to Alexa Fluor was then diluted in 3% BSA–PBS-T, the mixture was added to the cells, and the cells were incubated for 45 min. Mouse anti-IE1 was labeled with a Xenon 488 system per the manufacturer's protocol (Life Technology) and then incubated with the cells for 1 h at RT. Coverslips were put onto glass slides in ProLong Gold antifade reagent (Invitrogen). Morphology was quantified at a magnification of ×200 by counting 200 IE1-positive cells for lamin A/C morphology. Images were obtained at a magnification of ×400 on a Nikon Eclipse TS100 inverted microscope (Nikon).

Statistical analysis.

The Student t test was used to compare two samples, and we determined that the variances were not different using an F test prior to using a t test. Analysis of variance (ANOVA) was used to compare more than two samples, and the Breslow-Day/Tarone test of heterogeneity was used for the microscopic comparison of nuclear morphology. A significant P value (P < 0.05) is indicated by an asterisk in the figures, while nonspecific P values are indicated by the lack of an asterisk. For multiple comparisons, P values were adjusted using the Holm-Šídák post hoc test.

RESULTS

Cellular p21^Cip1 expression contributes to pUL27-mediated cell cycle arrest.

Mutations in the UL27 gene are selected for by passaging HCMV in the presence of a pUL97 kinase inhibitor, maribavir (MBV), or in the absence of the UL97 gene (22). In the absence of infection, expression of pUL27 results in increased levels of the CDK inhibitor p21^Cip1 and cell cycle arrest (28). These activities have been associated with the maximal antiviral activity of MBV and have led to the hypothesis that pUL27 acts in an antagonistic manner to pUL97 functions during infection (28, 54). To begin to define a potential relationship, we evaluated MBV-selected resistance mutations in UL27 for their ability to induce p21^Cip1 expression. UL27 encodes a 608-amino-acid protein, and diverse mutations have been identified within UL27, including the amino acid substitution R233S and a stop codon at C415 (C415stop) (22, 25, 27, 29). These changes occur throughout pUL27, suggesting that the absence of pUL27 activity may be due to altered protein structure. To assess p21^Cip1 expression, we transfected U-2 OS osteosarcoma (U2OS) cells with wild-type UL27 containing a green fluorescent protein (GFP) tag or MBV resistance-associated mutations R223S and C415stop. These mutations were selected for in the presence of MBV during infection (25). Expression of wild-type pUL27-GFP resulted in increased steady-state levels of the p21^Cip1 protein compared to those in the vector control (Fig. 1A). In contrast, expression of mutant pUL27-GFP with the R223S and C415stop mutations failed to increase p21^Cip1 levels (Fig. 1A). These data confirm that pUL27 mediates the induction of p21^Cip1 expression and demonstrate that UL27 mutants associated with MBV resistance fail to alter p21^Cip1 levels.

FIG 1.

Cellular p21^Cip1 expression contributes to pUL27-mediated cell cycle arrest. (A) Expression vectors for wild-type UL27 (wt) containing a GFP tag or MBV-resistant mutants (the C415stop [C415stp] and R223S mutants) along with an empty vector control (v) were transfected into subconfluent U2OS cells. Steady-state expression levels were evaluated by WB analysis using the indicated antibodies. (B and C) U2OS cells were grown in medium containing 0.5% serum for 24 h and then transfected with a control siRNA or an siRNA targeting p21^Cip1. After 48 h, the cells were transfected with either the empty vector or a pUL27-HA expression vector (UL27-HA). Samples were evaluated at 72 h by Western blotting using the indicated antibodies (B) or were evaluated by flow cytometry analysis for DNA content (C). For flow cytometry, cells were released into 7% serum with 100 ng/ml nocodazole for 18 h and then fixed and stained using an HA-specific antibody and propidium iodide. The propidium iodide levels in both pUL27-HA-negative and -positive cells were determined, and the data were analyzed using FlowJo software. (D) The data represent the cellular DNA content from three biological replicate experiments, with error bars showing standard deviations from the mean. Significance was determined by comparison of the data within each cell cycle phase by one-way ANOVA (*, P < 0.05). (E) Human foreskin fibroblasts were grown in 0.5% serum for 24 h and then released into 7% serum and infected with ADwt or ADd27 at an MOI of 1.0 IU/ml. Whole-cell lysates were evaluated by Western blotting using the indicated antibodies. Chemiluminescence was performed, and the results are graphed as the ratio of p21^Cip1 chemiluminescence/GAPDH chemiluminescence. Error bars represent standard deviations from three biological replicate experiments. Mock, mock-infected cells; Scram, scrambled siRNA.

Expression of pUL27 results in a cell cycle arrest consistent with elevated p21^Cip1 levels (28). To determine the contribution of p21^Cip1 to the arrest, we transfected U2OS cells with an siRNA targeting p21^Cip1 or with a scrambled siRNA control. Transfection of the gene-specific siRNA resulted in a substantial drop in expression compared to that for the control (Fig. 1B), and this reduction was maintained following expression of pUL27-HA (Fig. 1B). With the control siRNA, pUL27-HA still induced p21^Cip1, whereas the empty vector did not (Fig. 1B). To evaluate the impact on the cell cycle, transfected cells were incubated in medium containing 7% serum and 100 nM nocodazole for 16 h to allow progression to G₂/M. Cells were stained with propidium iodide for determination of the DNA content and an antibody against HA and then evaluated by flow cytometry. The DNA content was determined from gated populations of pUL27-HA-positive and -negative cells (Fig. 1C). In HA-negative cells, approximately 70% of control siRNA- and p21^Cip1-specific siRNA-transfected cells accumulated in G₂/M, suggesting that transfection of p21^Cip1-specific siRNA does not alter cell cycling, though this does not exclude the possibility of potential off-target effects. In contrast, 27% of pUL27-HA-expressing control cells existed in G₂/M (Fig. 1D), with the majority of cells being in G₀/G₁. Upon a reduction of p21^Cip1, however, 51% of pUL27-HA-positive cells were found in G₂/M (Fig. 1D). These data suggest that the pUL27-mediated induction of p21^Cip1 contributes to the cell cycle arrest but disruption of p21^Cip1 expression does not fully restore cycling.

During infection, the steady-state levels of p21^Cip1 increase very early and remain elevated until approximately 24 hpi, after which time their levels dramatically decrease (47). Several viral proteins have been shown to influence p21^Cip1 expression, including IE1, IE2, pUL27, and pUL29/28 (28, 48, 49, 55). To evaluate the influence of pUL27 expression on p21^Cip1 during infection, growth-arrested human foreskin fibroblasts were infected at 1 IU/cell with wild-type strain AD169 (ADwt) or a mutant variant lacking the majority of the UL27 open reading frame (ADd27) (28). Expression of p21^Cip1 was evaluated at various times postinfection by Western blotting. We observed a consistent increase in p21^Cip1 expression at 2 hpi and decreases in expression of p21^Cip1 at 8 and 12 hpi in the absence of UL27 (Fig. 1E). Using semiquantitative chemiluminescence, we estimated the differences in p21^Cip1 steady-state levels from three biological replicate experiments. In the absence of UL27, there was an estimated 2.7-fold increase in the p21^Cip1 level at 2 hpi, a 1.7-fold decrease in the p21^Cip1 level at 8 hpi, and a 7.0-fold decrease in the p21^Cip1 level at 12 hpi (Fig. 1E). These data suggest that pUL27 does contribute to regulating p21^Cip1 levels during infection.

HCMV pUL97 influences pUL27 phosphorylation and function.

pUL27 contains consensus sites for phosphorylation by the pUL97 kinase (S/T-X₄-R/L), suggesting that pUL27 is a target of the kinase (31, 56). To address this possibility, we began by investigating whether the two proteins could associate. First, we assessed a possible interaction within infected cells. For these studies, we used U373 astrocytoma cells, which support HCMV lytic replication and virus production and have higher transfection efficiencies than primary fibroblasts (10, 57, 58). As shown in Fig. 2A, cells were transfected with the empty expression vector (lanes 1 and 3) or an expression vector carrying pUL97-HA (lanes 2 and 4 to 7). The cells were then mock infected (lane 1) or infected with ADwt (lane 2) or a recombinant virus containing FLAG-tagged UL27 (AD27F) (lanes 3 to 7) (28). We evaluated interactions by immunoprecipitation using an anti-FLAG antibody and Western blot analysis. During infection, we detected low levels of both exogenously expressed pUL97-HA (Fig. 2A, lanes 6 and 7) and endogenous pUL97 (Fig. 2A, lane 3). pUL97 was not detected in the absence of the FLAG epitope in pUL27, suggesting that pUL97 was not nonspecifically precipitated (Fig. 2A, lane 2). Furthermore, no signal was detected using the HA antibody during infection with AD27F in the absence of pUL97-HA expression (Fig. 2A, lane 3). We then assessed the impact of MBV treatment on the association of pUL97 with pUL27 and found that kinase inhibition did not alter the association (Fig. 2B). These data suggest that pUL27 and the viral kinase pUL97 might associate during HCMV infection. However, this association represents a minor portion of the total pUL97 compared to the amount of pUL97 in the lysate, suggesting that only a small percentage of pUL97 associates with pUL27. Alternatively, the association might be transient, consistent with the observation that many kinases transiently associate with their substrates.

FIG 2.

HCMV pUL97 associates with pUL27 during infection and alters p21^Cip1 expression. (A) U373 cells were transfected with the empty vector or a pUL97-HA (+) expression vector or were untransfected (−). Cells were mock infected (lane m) or infected at an MOI of 1 IU/cell with ADwt or AD27F and collected at the indicated times. IP was performed with a FLAG-specific antibody, and WB analysis was performed using the indicated antibodies. (B) Cells were infected with AD27F, treated with 10 μM MBV (lanes +) or vehicle control (lanes −), and analyzed as described above. (C) Expression vectors for wild-type pUL27-GFP (wt), MBV-resistant mutants (a mutant containing only pUL27-GFP aa 1 to 414 [1-414] or the R223S mutant), or pUL97-HA were transfected into U2OS cells. Samples were immunoprecipitated using a GFP-specific antibody and evaluated by Western blotting using the indicated antibodies. (D) U2OS cells were transfected with empty plasmids or plasmids carrying UL27-HA, the UL97-HA wild type, or kinase activity-deficient UL97-HA (KD) and analyzed by Western blotting using the indicated antibodies.

We next evaluated the association of pUL27 and pUL97 and the functional impact on pUL27-mediated expression of p21^Cip1. Wild-type pUL27-GFP or the C415stop mutant was expressed in U2OS cells in the presence or absence of pUL97-HA. We found that both wild-type pUL27-GFP and the pUL27-GFP C415stop mutant still interacted with pUL97 (Fig. 2C, lanes 3 and 5), suggesting that the association involves the amino terminus of pUL27. Expression of pUL27-GFP with pUL97-HA resulted in reduced p21^Cip1 levels compared with the levels obtained by expression of pUL27-GFP alone (Fig. 2C, lanes 2 and 3), while the C415stop mutant had little impact on p21^Cip1 levels, regardless of kinase expression (Fig. 2C, lanes 4 and 5). We sought to confirm this observation using a different UL27 expression vector as well as the UL97K355M-HA kinase-dead expression vector carrying pUL97-HA containing the substitution K355M. We transfected U2OS cells with the pUL27-HA expression vector and either wild-type or kinase-inactive pUL97-HA containing the substitution K355M (59). Expression of pUL27-HA but not that of the vector or pUL97-HA alone resulted in an increase in p21^Cip1 levels (Fig. 2D, lanes 1 to 3). Upon coexpression of pUL27-HA and pUL97-HA, only a minor change in p21^Cip1 levels was detected (Fig. 2D, lane 4). In contrast, pUL27-mediated p21^Cip1 expression occurred in the presence of the kinase-deficient mutant (Fig. 2D, lane 6). These data demonstrate that expression of active pUL97 kinase disrupts the pUL27-mediated induction of p21^Cip1.

We observed changes in pUL27 mobility upon coexpression with wild-type but not inactive pUL97, suggesting changes in pUL27 modifications (Fig. 2D, lanes 4 and 6). Next, we asked whether pUL97 could alter the phosphorylation state of pUL27. We transfected U2OS cells using a pUL27-HA expression vector in the presence or absence of pUL97-HA (Fig. 3A). We immunoprecipitated pUL27-HA and evaluated changes in pUL27-HA phosphorylation (Fig. 3A). We evaluated phosphorylation using the phosphoprotein stain Pro-Q Diamond as well as antibodies against phosphoserine and phosphothreonine. Coexpression of pUL27-HA with pUL97-HA resulted in staining by both Pro-Q Diamond and the antiphosphoserine/antiphosphothreonine antibodies of a protein running at the predicted molecular mass of pUL27-HA (70 kDa) (Fig. 3A). This did not occur in the absence of the kinase even when similar levels of pUL27-HA were detected (Fig. 3A). Again, in these experiments we observed a pUL27-mediated increase in p21^Cip1 levels in the absence of pUL97 which were reduced upon pUL97 expression (Fig. 3A). Our data support the conclusion that pUL27 is a target of pUL97 and expression of active kinase antagonizes the pUL27-mediated regulation of cellular p21^Cip1.

We next evaluated the phosphorylation of pUL27 during infection. Fibroblasts were infected at 1.0 IU/ml with the AD169 virus as a control or with the FLAG-tagged UL27 recombinant virus in the presence of absence of 10 μM MBV. We immunoprecipitated pUL27-FLAG with a FLAG-specific antibody and evaluated the changes in phosphorylation using the phosphoprotein stain Pro-Q Diamond (Fig. 3B). As a control, GAPDH was observed in the lysate but not in the immunoprecipitated samples. Staining with a phosphoserine/phosphothreonine antibody was not included due to the nonspecific staining observed under infection conditions. We observed a band at 70 kDa, corresponding to the predicted molecular mass of pUL27, in both the untreated and MBV-treated lanes (Fig. 3B). We observed decreased Pro-Q Diamond staining of the 70-kDa band under MBV treatment conditions compared to the amount of staining achieved with the vehicle control (Fig. 3B), suggesting that MBV treatment decreases the level of pUL27 phosphorylation during infection. The remaining signal suggests that pUL27 may be phosphorylated by other kinases during infection with MBV.

Finally, using phosphoenrichment and mass spectrometry techniques, we explored the specific phosphorylation sites that occur on pUL27 (Fig. 3C). U2OS cells were transfected with the pUL27-HA and the pUL97-HA or UL97K355M-HA kinase-dead expression vectors. pUL27-HA and pUL97-HA were isolated using immunoprecipitation with an anti-HA antibody. Immunoprecipitate was digested using trypsin, and the sample was enriched for phosphopeptides using TiO₂. Sites of phosphorylation were identified using high-mass-accuracy mass spectrometry as described by Bigley et al. (19). We identified multiple sites of phosphorylation on pUL27-HA (Fig. 3C; see also Fig. S1 in the supplemental material) as well as on pUL97-HA (Fig. 3D; see also Fig. S2 in the supplemental material). We did not identify phosphopeptides from pUL27-HA when UL97K355M-HA was coexpressed. We used a similar approach to evaluate the phosphorylation of pUL27 during infection. Fibroblasts were infected with 3.0 IU/ml of wild-type virus or FLAG-tagged UL27 recombinant virus in the presence or absence of MBV and collected at 48 hpi. pUL27-FLAG was isolated by immunoprecipitation. We confirmed the phosphorylation of pUL27 at threonine 78 during infection using phosphopeptide enrichment and mass spectrometry (Fig. 3C). Phosphorylation at additional sites was not observed, and threonine 78 may represent the dominant phosphorylation site during infection. We did not identify pUL27 phosphopeptides during infection when we treated the cells with MBV. These data identify specific sites of phosphorylation occurring on pUL27 and demonstrate that pUL97 promotes the phosphorylation of pUL27.

Serum levels modulate UL27-deficient virus replication upon pUL97 kinase inhibition by MBV.

Reducing the amount of serum in the culture medium usually results in a G₀/G₁-specific arrest, decreased CDK activity, and increased steady-state levels of p21^Cip1 (60–62). In addition, other laboratories have demonstrated that reduced serum levels negatively impact HCMV replication yet increase the inhibitory activity of MBV (16). We next asked whether serum levels would alter the sensitivity of a UL27-deficient virus to MBV and kinase inhibition. For these experiments, we grew primary fibroblasts in 0.1% serum for 24 h, after which 89% of cells were in G₀/G₁, 4.8% of cells were in S, and 5.9% of cells were in G₂/M (Fig. 4D). We then completed the infections with ADwt or ADd27 virus at 1.0 IU/cell. We maintained infection in 0.1% serum or released the cells from starvation by placing them into 7.0% serum and evaluated the viral titers at 96 hpi. In untreated cells, we observed that the titers of both ADwt and ADd27 were reduced when cells were replicating in 0.1% serum compared to the titers obtained when cells were replicating in 7% serum (Fig. 4A). We repeated these experiments using 10 μM MBV to block kinase activity or the vehicle control and evaluated the viral titers. The results are represented as a relative ratio of the viral titers obtained from MBV-treated infected cells versus the viral titers obtained from vehicle control-treated (MBV-untreated) infected cells for each condition. MBV treatment resulted in a 1.5-log reduction in the amount of ADwt in 7% serum, whereas it resulted in a 2.2-log reduction in 0.1% serum (Fig. 4B). For ADd27 titers, we detected a 0.85-log reduction in 7% serum and a 2.0-log reduction in 0.1% serum under MBV treatment (Fig. 4B). The difference in sensitivity to kinase inhibition with 0.1% serum was not statistically significantly different between the wild-type virus and the UL27-deficient virus. These data confirm the impact of serum levels on MBV activity and demonstrate that the resistance phenotype of the UL27-deficient virus is nearly negated under low-serum conditions.

FIG 4.

Low serum levels rescue maribavir-mediated inhibition of a UL27-deficient virus. (A) Human foreskin fibroblasts were serum starved for 24 h in 0.1% serum. Cells were maintained in 0.1% serum or released into 7% serum and infected using 1 IU/cell of ADwt or ADd27. Viral titers, presented as the number of infectious units per milliliter of medium, were determined from cell-free virus at 96 hpi. (B) Cells were infected as described in the legend to panel A and treated with vehicle or 10 μM MBV. Viral titers were determined from cell-free virus at 96 hpi, and data are presented as the ratio of the virus titer for MBV-treated cells to the titer for MBV-untreated cells for each condition. Error bars represent standard deviations from two biological replicates, and significance was determined by one-way ANOVA (*, P < 0.05). (C) Steady-state protein levels were determined at 48 or 72 hpi by WB analysis using the indicated antibodies. c, control. (D) Cells were starved for 24 h in 0.1% serum and then fixed and stained using propidium iodide. The propidium iodide levels were determined and analyzed using FlowJo software. The data represent the cellular DNA content from two biological replicates. Error bars show standard deviations from the mean.

We next evaluated changes in the steady-state levels of p21^Cip1 as well as viral proteins during infection with ADwt or ADd27 under the various conditions. Using 0.1% serum, addition of 10 μM MBV resulted in slight increases in p21^Cip1 levels compared to those for the vehicle-treated controls for both ADwt and ADd27 at 48 and 72 hpi (Fig. 4C, left). In contrast, inhibition of the kinase in 7% serum resulted in increased p21^Cip1 levels in ADwt infection but not ADd27 infection at 48 and 72 hpi (Fig. 4C, right). We evaluated the differences in viral protein expression. The addition of MBV in 0.1% serum resulted in reduced levels of HCMV pUL44 and pp65 for both ADwt and ADd27 compared to those obtained by vehicle treatment (Fig. 4C, left). These reductions were also observed when using 7% serum, but we observed slight differences in pUL44 and pp65 expression between ADwt and ADd27 in the presence of MBV (Fig. 4C, right). The experiments were performed separately, and comparisons are limited to comparisons within each Western blot. Overall, our data demonstrate that the MBV resistance of the UL27-deficient virus is dependent upon serum levels and correlates with reduced levels of cellular p21^Cip1.

Levels of p21^Cip1 influence HCMV replication and susceptibility to kinase inhibition.

Because the levels of p21^Cip1 change during infection, we next asked whether the manipulation of p21^Cip1 would influence infection and MBV antiviral activity. To reduce p21^Cip1 levels prior to infection, we transfected serum-starved U373 cells with an siRNA against p21^Cip1 or scrambled control siRNA and maintained the cells in 0.5% serum. Western blot analysis showed a substantial reduction in p21^Cip1 levels at 24 h posttransfection (Fig. 5A). We infected the cells with ADwt and observed similar levels of IE1, pUL44, or pp65 between the different conditions at 96 hpi in U373 cells (Fig. 5B). Next, we quantified the impact of reducing p21^Cip1 levels prior to infection on viral replication in the presence or absence of kinase inhibition. siRNA-transfected cells were infected with ADwt or ADd27 and treated with 10 μM MBV or the vehicle as a control. We quantified the changes in virus release after 144 hpi. Disruption of p21^Cip1 expression resulted in a minor decrease in viral titers for both ADwt and ADd27 in vehicle-treated samples (Fig. 5C, left). Following kinase inhibition in samples transfected with the scrambled control, we quantified a 0.94-log reduction in the amount of ADwt virus and a 0.39-log reduction in the amount of ADd27 (Fig. 5C, right). In contrast, disruption of p21^Cip1 expression resulted in a 0.31-log reduction for ADwt and a 0.19-log reduction for ADd27 (Fig. 5C, right). Both viruses are less sensitive to MBV in U373 cells than in fibroblasts, which is likely due to the importance of cell culture conditions and cell type to MBV antiviral activity (63). The larger reduction in the sensitivity of ADwt than that of ADd27 upon a reduction of p21^Cip1 levels supports the conclusion that p21^Cip1 expression contributes to full MBV antiviral activity during infection.

FIG 5.

Manipulation of p21^Cip1 levels alters infection and maribavir-mediated inhibition. (A) U373 cells were serum starved for 24 h in 0.5% serum and then transfected with scrambled or p21^Cip1-specific siRNA for an additional 24 h in 0.5% serum. Protein expression was determined by WB analysis using the indicated antibodies. (B) At 24 h posttransfection, cells were infected at an MOI of 1 with ADwt or ADd27 virus in 7% serum. Viral protein expression was evaluated at 96 hpi using the indicated antibodies. (C) Infected cells were treated with vehicle (−) or 10 μM MBV (+). Viral titers were determined from cell-free virus at 96 hpi, and data are presented as the ratio of the virus titer for MBV-treated cells to the titer for MBV-untreated cells for each condition. Error bars represent standard deviations from four biological replicates, and significance was determined by one-way ANOVA (*, P < 0.05). (D) U373 cells were serum starved for 24 h in 0.5% serum and then transfected with the empty expression vector or the p21^Cip1-FLAG expression vector. Protein levels were determined by Western blotting using the indicated antibodies. (E) At 24 h posttransfection, cells were infected as described in the legend to panel B. (F) The impact of p21^Cip1 expression on MBV activity was evaluated as described in the legend to panel C. (G) Viral genome levels relative to the amount of cellular DNA at 96 hpi were determined using quantitative PCR analysis, and data are presented as the ratio of the viral genome level in MBV-treated cells to the level in MBV-untreated cells for each condition. Error bars represent standard deviations from four biological replicates, and significance was determined by one-way ANOVA (*, P < 0.05).

We were next interested in testing whether overexpression of p21^Cip1 could influence MBV antiviral activity and possibly restore the sensitivity of a UL27-deficient virus. For these studies, we transfected serum-starved U373 cells with a p21^Cip1 expression vector containing the FLAG epitope. We observed p21^Cip1-FLAG expression as well as increased total p21^Cip1 levels compared to that for the vector control at 24 h posttransfection (Fig. 5D). We infected cells with ADwt and observed no differences in steady-state IE1, UL44, or pp65 levels at 96 hpi, while p21^Cip1-FLAG levels remained elevated (Fig. 5E). At 144 hpi, we quantified 0.50- and 0.39-log reductions in the titers of ADwt and ADd27, respectively, when p21^Cip1-FLAG was expressed (Fig. 5F, left). Kinase inhibition of ADwt resulted in a 0.99-log reduction in viral titer under control conditions and a 0.78-log reduction when p21^Cip1-FLAG was expressed (Fig. 5F, right). The UL27-deficient virus displayed a 0.37-log decrease in viral titer under control conditions and a 0.77-log reduction when p21^Cip1-FLAG was expressed (Fig. 5F, right). At 96 hpi, we quantified a 0.69-log reduction in vDNA levels for ADwt when treated with MBV compared to the vDNA level for the untreated control, while MBV treatment of ADd27 resulted in a 0.22-log reduction (Fig. 5G). We observed similar reductions in vDNA levels when p21^Cip1-FLAG was expressed during ADwt or ADd27 infection (0.75 and 0.81 log units, respectively), and MBV treatment of ADwt and ADd27 did not result in a further reduction of vDNA levels (Fig. 5G). Neither p21^Cip1 knockdown nor overexpression altered viral entry, as evaluated by pp65 entry into the nucleus at 2 hpi (data not shown). Overall, these results indicate that p21^Cip1 expression is necessary for MBV antiviral activity and that maintaining p21^Cip1 protein levels during infection increases MBV sensitivity in virus with a UL27 gene deletion.

Impact of manipulating cellular CDKs on wild-type and UL27-deficient virus replication.

CDK1 and CDK2 activities increase as HCMV infection progresses (35, 44). This results from increased expression of the CDK proteins, their associated cyclins, and CDC25 phosphatases, as well as decreased levels of inactivating proteins Myt1, Wee1, and p21^Cip1 (44). CDK inhibitors have been demonstrated to increase MBV antiviral activity, likely due to the shared activities of the cellular CDKs and the viral CDK-like kinase pUL97 (9, 13). We speculated that UL27 mutations might give rise to MBV resistance, in part because p21^Cip1 levels are lower in UL27-deficient virus under MBV treatment conditions (Fig. 4C, right). Lower levels of p21^Cip1 are associated with increased CDK activity, which might complement the loss of pUL97 activity. To test this possibility, we infected fibroblasts with 1 IU/cell of ADwt or ADd27 virus under conditions of UL97 inhibition alone or in combination with CDK inhibition. First, we treated the infection with 10 μM MBV, 5 μM roscovitine, MBV and roscovitine, or the DMSO vehicle control. At 96 hpi, cell-free virus was collected and viral titers were determined. Treatment with 5 μM roscovitine resulted in a minimal decrease in viral yield for both viruses (Fig. 6A, left). Other labs have observed that higher concentrations of roscovitine can inhibit HCMV replication (37, 46). However, roscovitine has been shown to inhibit extracellular signal-regulated kinase 1 (ERK1) and ERK2 when used at higher concentrations (64, 65). Therefore, we elected to use a lower concentration of roscovitine. We assessed vDNA synthesis and found that ADd27 was less sensitive to MBV treatment than ADwt. Viral DNA was reduced 0.37 log unit under MBV treatment for ADwt and 0.21 log unit under MBV treatment for ADd27 (Fig. 6A, middle). Treatment of ADwt and ADd27 with roscovitine or with the combination of MBV and roscovitine resulted in a reduction in vDNA levels similar to that achieved with MBV treatment alone (Fig. 6A, middle). Treatment of ADwt with MBV resulted in a 1.6-log decrease in the viral titer, while treatment with MBV and roscovitine together resulted in a 2.1-log decrease (Fig. 6A, right). The titer of ADd27 was reduced 0.79 log unit under MBV treatment conditions and 1.8 log units under conditions of combined treatment with MBV and roscovitine (Fig. 6A, right). Cotreatment with MBV and roscovitine resulted in a greater reduction of the titer for the virus with the UL27 deletion than in that for the wild type compared to the reduction achieved with MBV treatment alone. These data indicate that CDK activity is important for the MBV resistance phenotype of the UL27-deficient virus.

FIG 6.

Inhibitors of CDK restore MBV antiviral activity in the absence of UL27. (A) Human foreskin fibroblasts were serum starved for 24 h and infected at 1 IU/cell with ADwt or ADd27 in 7% serum. Infected cells were treated with vehicle (−) or 5 μM roscovitine (Rosco) in the presence or absence of 10 μM MBV. Viral genome levels relative to the amount of cellular DNA were determined at 96 hpi using quantitative PCR analysis, and data are presented as the ratio of the viral genome level in MBV-treated cells to the level in MBV-untreated cells for each condition. Error bars represent standard deviations from three biological replicates, and significance was determined by one-way ANOVA (*, P < 0.05). Viral titers were determined from cell-free virus at 96 hpi, and data are presented as the ratio of the virus titer for MBV-treated cells to the titer for MBV-untreated cells for each condition. Error bars represent standard deviations from five biological replicates, and significance was determined by one-way ANOVA (*, P < 0.05). (B) Analysis of treatment with 2.5 μM Cdc25 inhibitor (CDC25i) under the conditions described in the legend to panel A. Error bars represent standard deviations from three biological replicates, and significance was determined by one-way ANOVA (*, P < 0.05).

Phosphorylation of CDKs at threonine 14 and tyrosine 15 results in their inactivation. Cdc25 phosphatases dephosphorylate CDKs at these residues, which is necessary for CDKs to activate the cell cycle by phosphorylating subsequent target proteins (66). To provide additional evidence that CDKs play a role in MBV resistance in the absence of pUL27, we inhibited Cdc25 phosphatases. Addition of 5 μM Cdc25 inhibitor II has been shown to inhibit CDK1 and CDK2 kinase activities and alter cell cycling (67). When we treated infected cells with the Cdc25 inhibitor, we observed a 0.4-log decrease in viral titer for both ADwt and ADd27 (Fig. 6B, left). Treatment of ADwt infection with 10 μM MBV resulted in a 1.3-log decrease in the viral titer compared to that obtained with no treatment, whereas cotreatment with MBV and the Cdc25 inhibitor reduced the viral titer by 2.3 log units compared to that obtained by treatment with the Cdc25 inhibitor alone (Fig. 6B, right). Treatment of the ADd27 virus infection with MBV reduced the viral titer by 0.61 log unit, but cotreatment with the Cdc25 inhibitor plus MBV resulted in a 2.0-log reduction (Fig. 6B, right). Again, the UL27-deficient virus was more sensitive than the wild type to inhibition by both compounds. These data support the conclusion that Cdc25 and CDK activities are necessary for MBV resistance in a UL27-deficient virus.

We repeated this experiment using 120 nM the CDK2 selective inhibitor CDK2 inhibitor II. At this concentration, which was previously demonstrated to inhibit activity (68), we did not observe a significant impact on the viral titer with or without MBV treatment (Fig. 7A). We sought to verify these results with a higher concentration of CDK2 inhibitor II. We did not observe a significant impact on viral titer with or without MBV using 300 nM CDK2 inhibitor II (Fig. 7A). We next used a CDK1-selective inhibitor, 2-cyanoethyl alsterpaullone (alsterpaullone), at 75 nM (69). In agreement with the findings of previous studies, CDK1 inhibition did not greatly impact the titer of either virus (Fig. 7B) (70). In wild-type virus infection, MBV reduced the viral titer by 1.3 log units (Fig. 7B). Addition of alsterpaullone treatment to MBV treatment only minimally reduced the viral yield compared to that achieved by MBV treatment alone, resulting in a 1.5-log decrease in viral titer (Fig. 7B). In contrast, treatment of the ADd27 virus with MBV only resulted in a 0.44-log reduction in the viral titer, whereas treatment with alsterpaullone and MBV resulted in a 0.97-log reduction in the viral titer (Fig. 7B). To identify whether there was a dose-dependent effect of alsterpaullone on the viral titer, we repeated the experiment with 150 nM alsterpaullone. The combined treatment with the higher concentration of alsterpaullone and MBV resulted in a 2.5-log reduction in the viral titer in wild-type virus and a 2.1-log reduction in ADd27, demonstrating a dose-dependent effect (Fig. 7B). These data demonstrate that the UL27-deficient virus is more sensitive to pUL97 kinase inhibition when treated with a CDK1 inhibitor. The results suggest that CDK1 is important for the complete MBV resistance phenotype of a UL27-deficient virus.

FIG 7.

Manipulation of CDK1 alters UL27-dependent MBV antiviral activity. (A) Human foreskin fibroblasts were serum starved for 24 h and infected with ADwt or ADd27 at 1 IU/cell in 7% serum. Infected cells were treated with vehicle (−), 10 μM MBV, or 120 or 300 nM (high dose [hi]) CDK2 inhibitor II (CDK2i). Viral titers were determined from cell-free virus at 96 hpi, and data are presented as the ratio of the virus titer for MBV-treated cells to the titer for MBV-untreated cells for each condition. Error bars represent standard deviations from five (120 nM) or three (300 nM) biological replicates. Statistical significance was calculated by one-way ANOVA (*, P < 0.05). (B) Analysis of treatment with 75 or 150 nM 2-cyanoethyl alsterpaullone (Alst) under the conditions described in the legend to panel A. Error bars represent standard deviations from five (75 nM) or three (150 nM) biological replicates, and significance was determined by one-way ANOVA (*, P < 0.05). (C) Cells were transfected with the UL97-HA expression vector, and the impact of 10 μM MBV, 5 μM roscovitine (Rosco), 120 nM CDK2 inhibitor II (CDK2i), 75 nM 2-cyanoethyl alsterpaullone, or 2 μM olomoucine (Olo) on pUL97 autophosphorylation was evaluated by Western blotting analysis. (D) U373 cells were serum starved for 24 h and then transfected with an empty vector or with a vector expressing constitutively active CDK1-FLAG (CDK1AF). At 24 h posttransfection, cells were infected with 1 IU/cell of ADwt or ADd27 in the presence or absence of 10 μM MBV. Samples were analyzed as described in the legend to panel A. Standard deviations from three biological replicates are shown, and significance was calculated using one-way ANOVA (*, P < 0.05).

HCMV pUL97 autophosphorylation can be seen as a mobility shift on SDS-polyacrylamide gels in Western blot assays, and this is dependent on its own kinase activity (71). We used this assay as a control to test whether CDK inhibitors altered pUL97 autophosphorylation. While MBV treatment resulted in the loss of the slower-migrating bands (Fig. 7C), we did not observe the same change when we treated cells with various CDK kinase inhibitors at the concentrations used in our experiments (Fig. 7C). These data suggest that pUL97 retains activity in the presence of these CDK inhibitors. Additionally, treatment with roscovitine, CDK2 inhibitor II, or alsterpaullone did not significantly alter the amount of pp65-positive cells at 2 hpi, suggesting that translocation of the pp65 tegument protein to the nucleus is unaltered by these CDK inhibitors (data not shown).

Finally, we further evaluated the importance of CDK1 for MBV antiviral activities by increasing CDK1 activity. We asked whether maintaining CDK1 activity would overcome the negative effects of the increased p21^Cip1 levels observed during MBV treatment. For this experiment, U373 cells were serum starved in 0.5% serum for 24 h and then maintained in a low concentration of serum and transfected with a control vector or with a vector expressing a constitutively active CDK1 with a FLAG epitope (CDK1AF) (51). CDK1AF contains substitutions at threonine 14 and tyrosine 15, preventing inactivation. At 24 h posttransfection, we infected cells with 1 IU/cell in 7% serum. Expression of constitutively active CDK decreased the titers of ADwt and ADd27 (Fig. 7D) (34). pUL97 inhibition of ADwt reduced the viral titer by 0.94 log unit under control conditions, whereas the viral titer was reduced only 0.37 log unit under conditions with CDK1AF expression (Fig. 7D). The ADd27 virus displayed a 0.42-log decrease in titer when pUL97 was inhibited and a 0.17-log decrease when pUL97 was inhibited in the presence of CDK1AF (Fig. 7D). We conclude that the constitutive activity of CDK1 decreases the antiviral activity of MBV, and these data together suggest that CDK1 activity is important for MBV resistance in a UL27-deficient virus.

CDK inhibition restored an MBV-mediated block in nuclear morphology changes in the absence of pUL27.

pUL97 has multiple functions, including disruption of the nuclear lamina through the phosphorylation of lamin A/C (10, 12, 72). During HCMV infection, the nuclear morphology becomes deformed and gaps arise in the lamina, an event that requires pUL97 (10, 12). This activity contributes to nuclear egress and impacts viral titers (10, 12). CDKs, including CDK1, phosphorylate lamin A/C to disrupt the nuclear lamina during mitosis (73). We next asked whether the nuclear deformity is rescued under conditions of pUL97 inhibition in a UL27-deficient virus and whether CDK activity might contribute to this process. We infected fibroblasts with 1 IU/cell of ADwt or ADd27 virus and treated the cells with 10 μM MBV, 5 μM roscovitine, roscovitine and MBV, or the vehicle control. We evaluated lamin A/C staining at 96 hpi using fluorescence microscopy to detect changes in nuclear morphology (Fig. 8A). In uninfected cells, the nuclear morphology appeared to be circular or oval and was designated a normal morphology (Fig. 8B). It has been reported that infection with HCMV results in a nuclear deformity characterized by a curved nuclear morphology (12), and we designated this change deformed (Fig. 8B).

FIG 8.

pUL27- and kinase-dependent inhibition of HCMV mediate changes in nuclear morphology. (A) Human foreskin fibroblasts were serum starved for 24 h and infected at 1 IU/cell with ADwt or ADd27 in 7% serum. Infected cells were treated with vehicle (i.e., untreated [Unt]), 10 μM MBV, 5 μM roscovitine (Rosco), or both compounds. Cells were fixed and stained using antibodies to HCMV IE1 and lamin A/C (Lam A/C). (B) HCMV-mediated changes in the nuclear lamina morphology were classified as deformed or normal in IE1-postive cells. (C) The data represent the percentage of at least 200 IE1-positive cells with deformed nuclear morphology relative to normal morphology under each condition. (D) Model depicting the relationship among pUL27, p21^Cip1, pUL97, CDK1, and MBV treatment. pUL27 induces p21^Cip1 at immediate early times of infection. pUL97-mediated phosphorylation of pUL27 eventually disrupts p21^Cip1 expression. MBV inhibition of pUL97 results in active pUL27, sustained increased levels of p21^Cip1, and CDK inhibition. The result is the disruption of virus release. In the absence of active pUL27 and elevated p21^Cip1 levels, endogenous CDK1 partially complements the MBV-mediated loss of pUL97 activity, supporting virus release.

We quantified the changes in nuclear morphology in IE1-positive cells observed in Fig. 8A in accordance with the methods described in previous studies (10, 12). We observed nuclear deformity in 60.5 to 68% of ADwt- or ADd27-infected cells treated with vehicle control or roscovitine (Fig. 8C). Under MBV treatment, 16.5% of IE1-positive ADwt-infected cells demonstrated a deformity (Fig. 8C). MBV-treated cells infected with ADd27 showed an intermediate change, with 43% of IE1-positive cells having deformed nuclei. Treatment of ADwt-infected cells with roscovitine plus MBV resulted in a deformity in only 11% of the cells. Treatment of ADd27-infected cells with roscovitine plus MBV resulted in a deformity in 13.5% of the cells, which was similar to the percentage obtained with MBV treatment in wild-type infection. Overall, these observations suggest that the changes in the nuclear morphology of cells induced by UL27-deficient HCMV can occur during MBV inhibition of pUL97 activity when CDKs are active.

DISCUSSION

In this report, we provide evidence of an antagonistic relationship between the activities of the pUL97 kinase and a second HCMV protein, pUL27. We have summarized this relationship in a model presented in Fig. 8D. Furthermore, this relationship is a determinant in the antiviral activity of the HCMV kinase inhibitor MBV. Consistent with the findings of our previous studies (28), we observed that expression of pUL27 alone is sufficient to increase the steady-state levels of the CDK inhibitor p21^Cip1 (Fig. 1 to 3). The increased p21^Cip1 level is the result of pUL27-dependent degradation of the cellular Tip60 acetyltransferase and altered p21^Cip1 RNA levels (28). At early times during infection using a virus lacking the UL27 gene, we observed reduced levels of p21^Cip1 compared to those obtained from wild-type virus (Fig. 1E). The response was altered upon coexpression of pUL97 kinase, resulting in reduced levels of pUL27-mediated induction of p21^Cip1 (Fig. 2C and D). This antagonism was dependent upon kinase activity, since coexpression using a kinase-deficient pUL97 mutant maintained elevated p21^Cip1 levels (Fig. 2D). During infection, inhibition of kinase activity by MBV also resulted in elevated p21^Cip1 levels using wild-type virus but not the UL27-deficient virus (Fig. 4C). A limitation to our studies is the absence of an additional independently derived mutant virus. However, our previous studies demonstrated that the UL27-deficient virus exhibits the same reduced sensitivity to MBV as a UL27 start codon-deficient mutant (28). One potential mechanism of regulation involves the phosphorylation of pUL27. We observed that expression of pUL97 resulted in the increased phosphorylation of pUL27 (Fig. 3A) and this was reduced during infection with MBV (Fig. 3D). Using mass spectrometry, we identified several sites of pUL97-dependent phosphorylation in pUL27, including pUL97 consensus sites (Fig. 3B and C). Overall, our studies have demonstrated that pUL97 antagonizes pUL27-mediated regulation of p21^Cip1 levels during HCMV infection (Fig. 8D).

The idea that pUL27 is regulated by pUL97 during infection is supported by differences in the kinetics of expression observed in the quantitative proteomics studies performed by Weekes et al. (74). HCMV pUL27 is expressed with immediate early kinetics, while pUL97 is expressed with early late kinetics. However, the functional role of pUL27 during infection and why it is regulated remain unknown. Regulation of p21^Cip1 by pUL27 involves the proteasome-dependent degradation of Tip60, which is a transient event early during infection (28). Tip60 acetyltransferase regulates cellular transcription and participates in the activation of the DNA damage response (reviewed in reference 75). Interestingly, herpesvirus kinases, including pUL97, also modify Tip60 (76). The Epstein-Barr virus (EBV) kinase BGLF4 induces Tip60 phosphorylation, resulting in increased acetyltransferase activity, DNA damage response signaling, EBV gene expression, and vDNA synthesis. During HCMV infection, short hairpin RNA-mediated disruption of Tip60 results in a significant reduction in vDNA synthesis. Intriguingly, human papillomavirus E6 protein induces Tip60 degradation (77), while it also requires Tip60-mediated signaling and ATM induction to support keratinocyte differentiation-dependent vDNA synthesis (78). Similar to herpesvirus kinases, cyclin B/CDK1 also phosphorylates Tip60, resulting in increased activity (79). Future studies will determine how phosphorylation alters pUL27 regulation of Tip60 activities during HCMV infection.

To date, the importance of p21^Cip1 regulation during infection has not been thoroughly explored. Efficient onset of HCMV infection occurs during G₀/G₁, with viral proteins constructing a pseudo-cell cycle state consisting of markers for G₁, S, and M while inhibiting cellular DNA synthesis (reviewed in reference 32). Depending on its phosphorylation status, p21^Cip1 can promote CDK4 activity while inhibiting CDK1/2 (80), and it is likely that p21^Cip1 participates in defining the pseudo-cell cycle. Supporting the possibility, Zydek et al. (33) have demonstrated that checkpoint-dependent activation of p21^Cip1 in primary fibroblasts in S/G₂ overcomes the cyclin A-mediated block in immediate early gene expression. We found that artificially maintaining or depleting p21^Cip1 levels by transfection in U373 cells resulted in a modest decrease in wild-type virus replication (Fig. 5). These findings suggest that altering the normal kinetics of p21^Cip1 levels during infection is detrimental to viral replication. The significance of p21^Cip1 to infection is more evident upon inactivation of the viral kinase using the compound MBV. Depletion of p21^Cip1 during infection resulted in reduced sensitivity to MBV and increased titers of wild-type HCMV (Fig. 5C). In contrast, maintenance of p21^Cip1 levels throughout infection resulted in increased sensitivity to MBV and decreased titers of a UL27-deficient virus (Fig. 5F). Elevated p21^Cip1 levels inhibited vDNA synthesis irrespective of pUL97 kinase inhibition, suggesting additional inhibitory activities (Fig. 5G). Overall, our data indicate that MBV antiviral activity is affected by p21^Cip1 levels.

HCMV pUL97 has been designated a viral CDK-like kinase because it shares overlapping targets and functions with the cellular CDKs (11–13). While CDKs are sensitive to inhibition by p21^Cip1, Hume et al. (11) have demonstrated that pUL97 is less sensitive to p21^Cip1. During infection, CDK activity increases concurrently with vDNA synthesis and is maintained at increased levels thereafter (35). This correlates with increased steady-state levels of cyclin B and CDK1 as well as cyclin E and CDK2 (35, 37). Previous studies have demonstrated that CDK inhibitors decrease viral replication and enhance MBV antiviral activity (70). Our studies confirm this observation and implicate a role for CDK1 during infection that is detectable in the absence of pUL97 kinase activity (Fig. 7), which is represented in the model in Fig. 8D. Expression of a constitutively active CDK1 resulted in decreased sensitivity to MBV and increased viral replication, with the greatest change being seen for wild-type virus (Fig. 7D). In contrast, inhibition of CDK1 (Fig. 7B) or the upstream activator Cdc25 (Fig. 6B), but not CDK2 (Fig. 7A), resulted in increased sensitivity to MBV or decreased replication, with the greatest change being observed for the UL27-deficient virus. These differences were detectable at the level of viral titers but not vDNA synthesis (Fig. 6A). One function of pUL97 is disruption and deformation of the nuclear lamina (10, 12). This contributes to egress of the nucleocapsid from the nucleus (8, 81). Similar to the findings of studies by Hamirally et al. (12), we also observed that disruption of the nuclear lamina is largely dependent on pUL97 activity (Fig. 8A and C). The nuclear deformity indicates disruption of the nuclear lamina. Interestingly, the absence of pUL97 kinase activity and pUL27 was associated with the nuclear lamina deformity during infection, but this did not occur during CDK inhibition (Fig. 8A and C). Inhibition of pUL97 kinase activity resulted in increased levels of p21^Cip1, but only if pUL27 was expressed (Fig. 4C). Together, these data suggest that CDKs can complement some activities of pUL97 which are disrupted by elevated levels of the pUL27-dependent induction of p21^Cip1 (Fig. 8D). A similar model of MBV resistance was previously proposed by Kamil and Coen (54). Our studies further support the notion that CDK inhibitors (70) or p21^Cip1 activators (28) might be useful in combination with MBV to effectively inhibit HCMV infections.