Identification of Human Cytomegalovirus Genes Important for Biogenesis of the Cytoplasmic Virion Assembly Complex

ABSTRACT

Human cytomegalovirus (HCMV) has many effects on cells, including remodeling the cytoplasm to form the cytoplasmic virion assembly complex (cVAC), the site of final virion assembly. Viral tegument, envelope, and some nonstructural proteins localize to the cVAC, and cytoskeletal filaments radiate from a microtubule organizing center in the cVAC. The endoplasmic reticulum (ER)-to-Golgi intermediate compartment, Golgi apparatus, and trans-Golgi network form a ring that outlines the cVAC. The center of the cVAC ring is occupied by numerous vesicles that share properties with recycling endosomes. In prior studies, we described the three-dimensional structure and the extensive remodeling of the cytoplasm and shifts in organelle identity that occur during development of the cVAC. The objective of this work was to identify HCMV proteins that regulate cVAC biogenesis. Because the cVAC does not form in the absence of viral DNA synthesis, we employed HCMV-infected cells transfected with synthetic small interfering RNAs (siRNAs) that targeted 26 candidate early-late and late protein-coding genes required for efficient virus replication. We identified three HCMV genes (UL48, UL94, and UL103) whose silencing had major effects on cVAC development, including failure to form the Golgi ring and dispersal of markers of early and recycling endosomes. To confirm and extend the siRNA results, we constructed recombinant viruses in which pUL48 and pUL103 are fused with a regulatable protein destabilization domain (dd-FKBP). In the presence of a stabilizing ligand (Shield-1), the cVAC appeared to develop normally. In its absence, cVAC development was abrogated, verifying roles for pUL48 and pUL103 in cVAC biogenesis.

IMPORTANCE Human cytomegalovirus (HCMV) is an important human pathogen that causes disease and disability in immunocompromised individuals and in children infected before birth. Few drugs are available for treatment of HCMV infections. HCMV remodels the interior of infected cells to build a factory for assembling new infectious particles (virions), the cytoplasmic virion assembly complex (cVAC). Here, we identified three HCMV genes (UL48, UL94, and UL103) as important contributors to cVAC development. In addition, we found that mutant viruses that express an unstable form of the UL103 protein have defects in cVAC development and production of infectious virions and produce small plaques and intracellular virions with aberrant appearances. Of these, only the reduced production of infectious virions is not eliminated by chemically stabilizing the protein. In addition to identifying new functions for these HCMV genes, this work is a necessary prelude to developing novel antivirals that would block cVAC development.

INTRODUCTION

Herpesvirus virion assembly is a complex process (reviewed in references 1 and 2). A portion takes place in the nucleus, which is the site of viral DNA replication, formation and filling of the nucleocapsid with viral DNA, and acquisition of the initial tegument layer. The major pathway of virion maturation involves the nascent capsid budding through the inner nuclear membrane into the lumen of the nuclear membrane, thus acquiring the primary envelope. Subsequently, the virion “infects” the cytoplasm through the outer nuclear membrane, leading to loss of the primary envelope. Tegument components are added during movement through the cytoplasm to a vesicle into which the particle buds, acquiring its mature envelope. The vesicle housing the virion is transported to the cell surface, where the virion is released in an exocytic process.

Human cytomegalovirus (HCMV) induces cytopathic effects that include cytoplasmic and nuclear enlargement and development of characteristic cytomegalic nuclear and cytoplasmic inclusions (reviewed in reference 3). The large cytoplasmic inclusion corresponds to the cytoplasmic virion assembly complex (cVAC) (4). cVACs are large cylindrical structures, one of which is present per infected cell, even in multinucleate syncytia. Nuclei take on a kidney-like (reniform) shape as they bend partially around the cVAC (5). Electron microscopic evidence indicates that the cVAC is the site of final tegumentation and envelope acquisition (6–8). Viral tegument proteins, envelope proteins, and some nonstructural proteins localize to the cVAC, and cytoskeletal filaments appear to radiate from a microtubule organizing center in the cVAC (4, 9–11). The cVAC is relatively devoid of endoplasmic reticulum (ER) markers. The ER-to-Golgi intermediate compartment (ERGIC), Golgi apparatus, and trans-Golgi network (TGN) form a cylindrical ring that outlines the cVAC (4, 10, 12).

Tooze et al. (7) treated cells with soluble horseradish peroxidase (HRP), which is taken up by recycling endosomes, and found its reaction product of oxidized diaminobenzidine in cytoplasmic vesicles that contained what appeared to be tegumented and enveloped HCMV virions, providing direct evidence that the virion exit vesicle is related to recycling endosomes. Consistent with this, we and others found that molecular markers of recycling endosomes (early endosome antigen 1 [EEA1] and Rab11) stain vesicles that cluster at the center of the cVAC (10, 12–15). Exactly how virions mature in the cVAC remains a mystery, and it has not been demonstrated whether a fully formed cVAC is necessary for efficient virion production. Although structures morphologically similar to the HCMV cVAC have not been described for other herpesviruses, a recent study found that herpes simplex virus 1 virions acquire their envelope at tubular structures that share many properties with recycling endosomes (16).

Although the full set of viral regulators of cVAC biogenesis has not been identified, some things are known. (i) cVACs do not form in the absence of viral DNA synthesis (reference 11 and our unpublished observations), indicating that its formation is dependent on expression of one or more viral late genes; a gene(s) from an earlier kinetic class may also be involved. (ii) Treatment of cells with nocodazole, which depolymerizes microtubules, leads to rapidly reversible cVAC disruption (17). While microtubules are important (even essential) components of the cVAC, they are not likely to be regulators of cVAC biogenesis. (iii) The kinase activity of pUL97 is required for formation of perinuclear complexes that correspond to cVACs and for efficient production of infectious virions, but UL97 activity alone is insufficient for cVAC development (11, 18, 19). (iv) Although not examined in detail, images in papers describing some HCMV mutants that have defects at late stages of virion assembly are consistent with at least some features of cVAC-like structures forming in the absence of UL32, UL91, UL96, and UL99 (20–23). (v) In contrast, insertion of a stop codon into the UL103 open reading frame resulted in the absence of well-formed Golgi rings, but the recycling endosomal compartment was not studied (15). In addition, in cells infected with a pUL71-deficient virus, viral proteins that normally associate with the cVAC are mislocalized (21). (vi) cVACs form in cells infected with highly passaged laboratory strains of HCMV that lack some genes present in wild-type viruses (4, 5, 8–15, 17, 20–23), as well as after infection with a low-passage-number virus that carries a nearly complete complement of HCMV genes (24). Interestingly, an HCMV locus spanning UL133 to UL138 is required for cVAC development in human microvascular endothelial cells, but not in embryonic lung fibroblasts (24), indicating that cell-specific viral factors can be involved. (vii) In addition to protein regulators of cVAC biogenesis, the HCMV microRNAs (miRNAs) miR UL112-1, US5-1, and US5-2 target mRNAs of several host proteins involved in regulation of the cellular secretory apparatus (25). A recombinant virus that does not express these miRNAs did not induce formation of cVACs and produced significantly fewer infectious virions. Ectopic expression of these microRNAs had some effects on Golgi structures, but full cVACs did not form.

In prior studies, we described the three-dimensional (3D) structure of the cVAC and the extensive remodeling of the cytoplasm and shifts in organelle identity that occur during cVAC development (13). The objective of this work was to identify HCMV protein-coding genes that regulate cVAC biogenesis. Using small interfering RNAs (siRNAs) that target a set of 26 candidates, we identified three HCMV genes (UL48, UL94, and UL103) whose absence results in abrogation of cVAC development without affecting the expression of other viral genes. Our results expand the set of functions assigned to HCMV genes and will enable identification of the network of interactions between HCMV and the host cell secretory apparatus, provide new information about the process of virion assembly, and define new targets for development of novel antivirals.

MATERIALS AND METHODS

Cells and virus.

HCMV (AD169: American Type Culture Collection, Manassas, VA) was grown in low-passage-number human foreskin fibroblasts (HFFs) in complete Dulbecco's modified Eagle medium (DMEM) plus sodium pyruvate and minimum essential medium with nonessential amino acids (Invitrogen, Carlsbad, CA), GlutaMax (Invitrogen), and 5% fetal bovine serum (FBS) (HyClone, Logan, UT). The virus was titrated by plaque assay on low-passage-number diploid human foreskin fibroblasts. For growth curves, supernatant virus was collected, purified, and stored at −80°C. The siRNA experiments reported here were performed in low-passage-number human lung fibroblasts (HLF) (a gift from John Stewart, Centers for Disease Control and Prevention, Atlanta, GA) that were propagated in the medium described above but with 10% FBS and penicillin and streptomycin. Antibiotics were omitted during experiments.

siRNA screening and immunofluorescence assay (IFA).

Cocktails of four siRNAs that specifically target each candidate HCMV regulator of cVAC biogenesis were designed and synthesized by Dharmacon Inc. (Lafayette, CO). For the transfection experiments, 2 × 10⁴ HLF per well were seeded onto 0.2% gelatin-coated 8-well glass chamber slides (LabTek, Nunc, Rochester, NY; catalog number 177402) for 1 h at 37°C. The following day, the cells were transfected with 50 pmol Smartpool siRNA using Lipofectamine 2000; 6 h after transfection, the cells were infected at a multiplicity of infection (MOI) of 0.2 in DMEM containing 5% fetal bovine serum. As previously described (16), after 120 h, the cells were washed with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde in PBS (lacking Ca²⁺ and Mg²⁺) at pH 7.4, and then autofluorescence was quenched by incubation for 15 min in 50 mM ammonium chloride. The cells were subsequently permeabilized and then successively incubated with the primary antibodies (described in Table 1) and fluorescence-tagged secondary antibodies (Alexa Fluor 488-conjugated goat anti-mouse IgG and Alexa Fluor 568-conjugated goat anti-rabbit IgG; both from Molecular Probes, Carlsbad, CA). Mounting was done with Vectashield containing DAPI (4′,6-diamidino-2-phenylindole) (Vector Laboratories Inc., CA). Imaging was done on a Leica TCS SP5 laser scanning confocal microscope.

TABLE 1.

Antibodies used in IFA and immunoblot assays

Antibody target	Host/isotype, clone	Source, catalog no.
Epitope tag
V5 (14 amino acids)	Mouse monoclonal/IgG2aκ	Life Technologies, R960-25
Cellular
GAPDH (36 kDa)	Mouse monoclonal/IgG1, clone GA1R	ThermoScientific, MA5-15738
GM130 (130 kDa)	Mouse monoclonal/IgG1(κ), clone 35/GM130	BD Biosciences, 610822
EEA1 (180 kDa)	Rabbit polyclonal	Abcam, ab2900
HCMV
IE1 and IE2 (IE1, 72 kDa; IE2, 86 kDa)	Mouse monoclonal/IgG1(κ), clone CH160	Virusys, P1215
IE2 (86 kDa)	Mouse monoclonal, clone 8B1.2	Chemicon (Millipore), MAB810
pUL44 (CMV ICP36), 46 kDa	Mouse monoclonal/IgG1(κ), clone 10D8	Virusys, CA006-100
UL48 (N-terminal region)	Rabbit polyclonal	Wade Gibson (Johns Hopkins University) (39)
pUL99 (pp28)	Mouse monoclonal/IgG2A(κ), 5C3	Virusys, CA004-100
gB (55 kDa and 110 kDa)	Mouse monoclonal/IgG1(κ), clone 2F12	Virusys, CA005-100

Construction of recombinant viruses.

Recombinant HCMVs were generated by recombineering using pADcre, an AD169 bacterial artificial chromosome (BAC), in SW105 (a gift from Donald L. Court, National Cancer Institute), an Escherichia coli strain that contains a temperature-sensitive recombinase gene and an arabinose-inducible Flp gene (26, 27). pAD/Cre contains the AD169 genome flanked by loxP sites that excise the bacterial component of the BAC upon transfection into mammalian cells. Due to overlapping adjacent coding regions at the 5′ ends of the UL48 and UL103 genes, the FKBP segment (28) was inserted at the 3′ end of the target genes. Selective markers (GalK and kanamycin) flanked by FKBP and FRT sequences were PCR amplified from pYD-C630 (provided by Dong Yu, Washington University) (29). The primers for recombineering are listed in Table 2. Positive selection of the FKBP GalK/kanamycin (Kan) cassette was carried out on kanamycin-Luria broth agar plates. Selection markers were removed by inducing the SW105 Flp recombinase, leaving UL48-FKBP or UL103-FKBP with an FRT site in frame at the C terminus. Removal of the GalK selection marker was verified by selection on galactose indicator plates. Galactose-negative colonies were picked and verified by restriction enzyme analysis, PCR, and DNA sequencing. To create UL103-FKBP-V5, the GalK/kanamycin cassette was replaced by amplifying and recombining the V5 tag. Negative selection of GalK was performed on M63 minimal medium plates supplemented with glycerol, d-biotin, l-leucine, 2-deoxygalactose, and chloramphenicol. Loss of GalK was verified as described above. All viruses were reconstituted by using Lipofectamine 2000 to transfect HFFs with 8 to 10 μg of BAC DNA, followed by propagation in the presence of 1 μM Shield-1 (catalog no. CIP-AS1; Cheminpharma, Farmington, CT), which was replaced every 48 h to maintain its activity. Virus was collected when the monolayer showed >80% cytopathic effect. Virus stocks were grown by infecting HFFs at a multiplicity of infection of 0.01 in the presence of 2 μM Shield-1; extracellular virions were partially purified by centrifugation through a 20% sucrose cushion.

TABLE 2.

Primers used for generating recombinant viruses

Virus	Primer sequencea	Amplification target
UL48-FKBP	5′-CGCAATCCGTACAGGACACTATTCAACACATGCGGTTTCTCTATCTTTTGatgggagtgcaggtggaaaccatc-3′	FKBP-GalK/Kan tag
	5′-ACGATAAAAATCCTATTGTTTTTATTACCCGCTACTGTCAGTGTCGGTTAgctggagctccaccgcgggaagttc-3′
UL103-FKBP	5′-GTTGCGTGTTTTTTTTTTTTCTATGATATGCGTGTCTAGTTCGCTTCTCAgctggagctccaccgcgggaagttc-3′	FKBP-GalK/Kan tag
	5′-TGCCCTCACCCCCCAAGCTGCCGCCGCGCTGGGAACGAGGAGAGGAAGAGatgggagtgcaggtggaaaccatc-3′
UL103-FKBP-V5	5′-GTTGCGTGTTTTTTTTTTTTCTATGATATGCGTGTCTAGTTCGCTTCTCAcgtagaatcgagaccgagga-3′	V5 tag
	5′-CACATGCCACTCTCGTCTTCGATGTGGAGCTTCTAAAACCGGAAGAATTCggtaagcctatccctaaccc-3′

^aThe uppercase sequences correspond to the 50-bp segments of viral DNA needed for recombination into the viral genome. The lowercase sequences are regions needed to enable amplification of exogenous sequences (e.g., the V5 epitope) intended to be inserted into the BAC.

Immunoblots.

HFFs were grown to confluence in T25 flasks and mock infected or infected at an MOI of 0.1. At 5 days postinfection (p.i.), the cells were washed with PBS and then lysed with RIPA buffer (0.1 M HEPES, pH 7.4, 0.1% sodium deoxycholate, 150 mM NaCl, 1% NP-40, 0.1% SDS, and Roche protease inhibitor). Protein concentrations were determined by bicinchoninic acid (BCA) assay (Pierce, Rockford, IL). Equal amounts of protein solubilized in 2× SDS-Laemmli buffer with 2-mercaptoethanol were separated in 10% SDS-polyacrylamide gels, transferred to nitrocellulose membranes (Whatman, Florham Park, NJ), probed with primary antibodies, and then reacted with HRP-conjugated goat anti-rabbit/mouse IgG secondary antibodies (Thermo Scientific, Rockford, IL). Reactions were detected with the Supersignal West Pico Chemiluminescent substrate (Thermo Scientific, Rockford, IL) on autoradiography film (GE Healthcare, Pittsburgh, PA). The antibodies are listed in Table 1.

Electron microscopy.

Transmission electron microscopy was done by Hong Yi at the Robert P. Apkarian Integrated Electron Microscopy Core Facility of Emory University, Atlanta, GA, using methods developed Hong Yi in collaboration with J. Ahlqvist, R. Tandon, and E. Mocarski (15, 20). Briefly, HFFs were infected in the presence and absence of Shield-1 with the parental (pAD/Cre) or the UL103-FKBP virus at an MOI of 0.3. At 120 h p.i., the cells were washed with fixative (2.5% glutaraldehyde in 0.1 M sodium phosphate buffer [pH 7.4]) and then incubated in fixative for 20 min at room temperature. After overnight storage in fixative at 4°C, the cells were shipped submerged in the fixative. The cells were postfixed in the same buffer with 1% osmium tetroxide, dehydrated through a graded series of ethanol, and then embedded in an epoxy resin. The cell culture dishes were broken, and ultrathin sections were then cut and counterstained with uranyl acetate and lead citrate. Ultrathin sections were examined with a JEOL JEM-1400 or a Hitachi H-7500 transmission electron microscope operated at 80 or 75 kV. Images were collected using a SIA 12C or Gatan US1000 charge-coupled-device camera.

RESULTS

Screening strategy to identify viral genes required for cVAC development.

Based on the requirement for viral DNA synthesis for cVAC biogenesis and the hypothesized importance of the cVAC in virion maturation, we identified 26 HCMV genes as possible regulators of cVAC biogenesis (Table 3) and used siRNAs to individually silence them during infection. Effects on cVAC development were visualized by confocal microscopic analysis of markers that are collectively diagnostic for cVAC structure.

TABLE 3.

Targets for siRNA screening^a

Function	Replicationb	Gene(s)
Tegument	R	UL97c
	E	UL32, UL48, UL82, UL94, UL99
Glycoproteins	E	UL55, UL73, UL75, UL100, UL115
Unknown	R	UL21, UL29, UL30, UL69, UL103, UL117
	E	UL34, UL49, UL71c, UL76c, UL91, UL92, UL93, UL95, UL96

^aHCMV early-late and late genes essential or important for viral growth. The replication data are from references 40 and 41; kinetic class assignments are from reference 42.

^bR, reduced (viral growth reduced by at least 50-fold); E, essential (no viral replication in its absence).

^cKinetic class not assigned (42).

The synthetic siRNAs used were mixtures of 4 siRNA duplexes, each of which independently targets the transcript of interest (Dharmacon Smartpools). The use of such pools instead of single siRNAs increases specificity by reducing the input concentration of individual siRNAs, thereby reducing off-target effects. Pools also increase the likelihood of successful silencing, because four regions of the target gene are targeted at once. Although antibodies are not available against most of the proteins whose expression we targeted, similarly designed siRNA pools that target HCMV proteins for which we have antibodies (IE2, MDBP, UL99, US17, and US18) delivered satisfactory results for every target tested (reference 30 and data not shown). Nonetheless, negative effects of individual siRNA pools should not be taken as proof that the targeted gene has no effect on cVAC biogenesis.

Screening experiments were done in 8-well glass chamber slides. HLF were transfected with the siRNAs 6 to 8 h prior to infection with fully infectious HCMV (AD169). Note that siRNAs do not affect input protein levels but reduce the levels of proteins expressed from genes transcribed during infection. At 120 h p.i., the cells were fixed and stained with the indicated antibodies.

The antibodies used in our antibody cocktails give reliable and distinct staining patterns that are collectively diagnostic for the cells being infected, expression of viral late genes, and formation of the cVACs (Fig. 1A and Table 4). To validate the screening strategy, we used 100 μg/ml phosphonoacetic acid (PAA)-treated, HCMV-infected cells as a prototype of a gene-silencing experiment. PAA inhibits viral DNA synthesis, consequently inhibiting cVAC biogenesis in HCMV-infected cells. As shown in Fig. 1B, in the presence of PAA, markers of the secretory apparatus in infected cells resemble those in uninfected cells.

FIG 1.

Validation of the siRNA screening system. (A) Diagrammatic representation of the cVAC staining pattern in HCMV-infected human lung fibroblasts and the predicted outcomes following silencing of cVAC biogenesis regulators. The patterns are representations of previously published data (10, 12, 13). (B) Inhibition of cVAC biogenesis by PAA (catalog no. P6909; Sigma-Aldrich, St. Louis, MO) treatment in HCMV-infected fibroblasts; confocal images are shown alongside the predicted patterns.

TABLE 4.

Targets and purposes of antibody cocktails used in screening assays

Cocktail	Antibodiesa						Purposeb
	Mouse monoclonal				Rabbit polyclonal
	IE2	pUL99	EEA1	GM130	Mann II	US18
A	+				+		Confirms infection and Golgi ring formation
B		+			+		Confirms Golgi ring formation and late-gene expression and localization
C			+		+		Confirms Golgi ring formation and EEA1 relocalization
D				+	+		Confirms Golgi ring formation
E	+					+	Confirms infection and late-gene expression and localization
F		+				+	Confirms late-gene expression
G			+			+	Confirms EEA1 relocalization and late-gene expression and localization
H				+		+	Confirms Golgi ring formation and late-gene expression and localization

^a+, present in cocktail.

^bIE2, pUL99, and US18 detect different stages of infection; GM130 and Mann II stain Golgi rings; and EEA1 localizes at the center of cVAC.

HCMV genes involved in cVAC biogenesis identified through the use of siRNAs.

Over 400 confocal images were analyzed as part of the screening process. Phenotypes observed after siRNA treatment included no effect, toxic effects, absence of HCMV late-gene expression, and inhibition of cVAC formation in cells where HCMV late genes are expressed. Candidate regulators of cVAC biogenesis were defined as genes whose silencing did not affect HCMV gene expression but prevented formation of perinuclear Golgi rings surrounding concentrations of EEA1 staining, with emphasis on the EEA1 patterns.

The results from our siRNA screen of 26 HCMV genes are summarized in Table 5. Some siRNAs had no visible effects on infected cells, while others caused extensive cytopathic effects, including blobs of dead cell nuclei, small nuclei, and irregularly shaped nuclei. These diverse effects might be products of silencing viral genes that are involved in regulating any of a variety of cellular processes, including protein secretory pathways, cellular metabolism, and antiapoptotic activity.

TABLE 5.

Properties and silencing results of HCMV genes screened by siRNA

Gene	Gene propertiesa	siRNA results			Candidate cVAC regulatorc
		Unusual cytopathic effectsb	Late-gene expression	Effect on Golgi markers and EEA1
UL21A	Enhances replication	Yes	No change in expression of pUS18/gB	Small effect on Golgi markers, but not on EEA1
UL29	Unknown	Yes (many nuclei are very small; blob of dead cells and empty patches)	No change in expression of pUS18 and pUL99	None
UL30	Unknown; alters PML morphology	Yes (blob of dead cells and empty patches)	No change in expression of pUS18/gB	Small effect on Golgi markers, but not on EEA1
UL32	Major tegument protein; binds to capsid	Yes (blob of dead cells and altered nuclear form)	Reduced (altered distribution of pUL99 [mainly cytoplasmic])	None
UL34	Represses US3 transcription; shuttles between nucleus and cytoplasm	Yes	Altered distributions of pUL18 and pUL99	Golgi markers did not form good rings, but EEA1 was still in tight perinuclear cluster.
UL48	Large tegument protein; ubiquitin-specific protease (N-terminal region); involved in capsid transport	Yes	Higher proportion of pUS18 in non-cVAC cytoplasm; expression of pUL99 reduced	Fuzzy and diffuse EEA1; altered localization of Golgi markers and absence of Golgi rings; few conventional cVACs	Yes
UL49	Unknown	Yes (many small nuclei)	No change in expression of pUS18, but gB was more diffuse.	Small effect on Golgi markers, but not on EEA1
UL55 (gB)	Virion envelope glycoprotein, gB; involved in heparan sulfate-mediated virion entry and cell-to-cell spread	Yes	Altered distribution of pUS18 and reduced gB expression	Small effect on Golgi markers, but none on EEA1
UL69	Tegument protein; multifunctional regulator; shuttles between nucleus and cytoplasm; inhibits pre-mRNA splicing; exports virus mRNA from nucleus	Yes (empty patches; irregular forms of nucleus and sometimes stretched)	Higher expression of pUS18 in non-cVAC cytoplasm; altered expression pattern of pUL99	None
UL71	Tegument protein; involved in virion morphogenesis	Yes (blobs of dead cells)	Reduced; altered distribution of pUL99	None
UL73 (gN)	Virion envelope glycoprotein, gN; complexed with envelope gM; supports virion morphogenesis	Yes (many small nuclei)	Expression of pUS18 reduced; altered distribution pattern of pUL99	None
UL75 (gH)	Virion envelope protein, gH; complexed with envelope gL; involved in cell entry and cell-to-cell spread	Yes (many small nuclei)	Expression of pUS18 reduced; altered distribution pattern of pUL99; mostly cytoplasmic	Reduced expression of Golgi markers and EEA1 but no changes in their arrangement
UL76	Virion protein; affects translation of UL77	Yes	Small altered expression of pUS18 and pUL99	Altered distribution of Golgi markers but not EEA1
UL82	Tegument phosphoprotein pp71; upper matrix protein; involved in gene regulation	Yes (many small nuclei)	No change in expression of pUS18, but reduced gB expression	Altered and reduced expression of Golgi markers, but not on EEA1
UL91	Unknown	Yes (many small nuclei)	No change in expression of pUS18 and pUL99	Altered distribution of Golgi markers but not EEA1
UL92	Unknown; members of this family found in several herpesviruses, including EBV BDLF4, HCMV UL92, HHV-8 ORF31, and HSV-6 U63	Yes (many nuclei are very small; blob of dead cells and empty patches)	Slightly altered expression of pUS18 and pUL99	None
UL93	Capsid associated; involved in DNA encapsidation and capsid transport	Yes (blob of dead cells and empty patches)	Slightly altered expression of pUS18, but not in pUL99	None
UL94	Tegument protein; involved in virion secondary envelopment	Yes	No effect on pUL99 or pUS18 expression levels.	Greatly reduced EEA1 expression; partial effect on Golgi markers, forming small Golgi ring	Yes
UL95	Late-gene activator; found in several herpesviruses, including EBV BGLF3 and other UL95 proteins (e.g., HCMV UL95, HSV-1 UL34, and HHV-6 U67)	Yes (blob of dead cells and empty patches)	Slightly altered expression of pUS18, but not pUL99	Altered distribution of Golgi markers, but not EEA1
UL96	Tegument protein; stabilization of nucleocapsids during nucleocytoplasmic translocation; possibly involved in virion morphogenesis	Yes (many blobs of dead cells)	Altered distribution and increased expression of pUL99	None
UL97	Tegument protein; virion serine/threonine protein kinase	Yes	Reduced	None
UL99	Myristylated tegument protein; involved in virion morphogenesis	Yes	Reduced	None
UL100 (gM)	Virion envelope glycoprotein, gM; 8 transmembrane domains; complexed with envelope gN; involved in virion morphogenesis	Yes (blob of dead cells with nuclear debris, patches of empty spaces and many small nuclei)	Slightly altered distribution of pUS18 and pUL99	Altered expression of Golgi markers, but not of EEA1
UL103	Tegument protein; involved in virion and dense-body egress; DNA encapsidation	Yes (big empty patches)	US18 looks normal, but altered distribution of UL99	Golgi markers and EEA1 are greatly reduced; altered distribution of Golgi markers	Yes
UL115	Virion glycoprotein gL; complexed with envelope gH; involved in cell entry; involved in cell-to-cell spread	Yes (blobs of dead cells)	Less abundant (late virion proteins/cVAC markers)	Reduced EEA1 expression; little effect on Golgi markers; Golgi rings with EEA1 center still present
UL117	Enhances replication compartment formation; impairs cellular DNA synthesis	Yes (many are very small and bright blue blobs [apoptotic debris?])	Expression of pUS18 alteredl fuzzier cVAC association; altered distribution pattern of pUL99 in some cells	Little effect on Golgi markers or EEA1

^aGene function information is reviewed in reference 43. EBV, Epstein-Barr virus; HHV-6/8, human herpesvirus 6/8; HSV-6, herpes simplex virus 6.

^bIn most instances, unusual cytopathic effects consisted of numerous blobs of dead cells and altered nuclear forms (smaller, elongated, or irregularly shaped).

^cCandidate cVAC regulators are the genes we selected for further study; this does not preclude other genes being important in the process.

The three genes whose siRNAs resulted in the most significant and specific effects on cVAC development were UL48, UL94, and UL103. In the presence of siRNAs against each of these genes, HCMV late genes were expressed; Golgi rings formed in a few cells; and the early/recycling endosome machinery was dispersed, nearly absent, or otherwise disrupted (Fig. 2). An important pathognomonic change associated with HCMV-infected cells is the development of enlarged nuclei that take on a reniform (kidney bean) shape; these changes are important during cVAC biogenesis (5). Knockdown of all three of the candidate cVAC biogenesis regulators (UL48, UL94, and UL103) resulted in significantly lower ratios of reniform to oval nuclei than in untreated cells and cells treated with siRNAs against several other genes targeted in our screening (Fig. 3). Some of the known properties of these candidate regulators are listed in Table 6.

FIG 2.

siRNA-based identification of UL48, UL94, and UL103 as candidate regulators of cVAC biogenesis. Cells were treated with siRNA pools designed to target the indicated HCMV genes, stained with antibodies against the indicated cellular markers, and then examined by confocal microscopy. A subset of the results obtained using the full panel of markers illustrated in Fig. 1 is shown. The staining patterns for the full set of markers were consistent with UL48, UL94, and UL103 playing important roles in cVAC biogenesis.

FIG 3.

Ratios of reniform and oval nuclei in HCMV-infected cells after gene silencing. HCMV-infected cells with reniform or oval nuclei were counted in cells with or without silencing of the indicated genes. The significance of the differences was evaluated using the chi-square test. n, number of cells examined for each condition; p, level of significance; ns, not significant.

TABLE 6.

Properties of candidate regulators of HCMV cVAC biogenesis^a

HCMV gene	Homolog			Functions and properties of HCMV homolog	Interacting viral partners (HCMV only)
	HSV	Gammaherpesvirus	HHV-6
UL48	UL36	ORF64	U31	Essential	pUL45, pUL47, pUL50, pUL69, pUL88, pUL103, pUL132
				Largest tegument protein
				Has deubiquitinating enzyme activity
				Role in intracellular capsid transport
UL94	UL16	ORF33	U65	Essential	pUS22, pUL72, pUL82, pUL99
				Tegument protein
				Binds single-stranded DNA
				Cytoplasmic egress
UL103	UL7	ORF42	U75	Elimination of pUL103 expression reduces virus replication by 102 to > 104-fold and results in reduced plaque size.	pUL22A, pUL48N, pUL103
				May have a role in viral replication
				Tegument protein
				Associated with the cVAC
				Virus particle and dense-body egress

^aThe information in this table is from references15, 37, 40, 41, and 44–48).

Use of regulated protein destabilization to verify siRNA screening results.

To independently confirm the siRNA results and to address specificity issues that might arise from overlapping transcripts previously described across UL48, UL94, and UL103, we employed regulated FKBP-mediated protein destabilization of the proteins of interest (Fig. 4) (28). The FKBP moiety targets tagged proteins for rapid proteasomal degradation, and the stability of the proteins tagged with FKBP domains can be regulated by the concentration of Shield-1. Due to coding region overlaps between the genes of interest and their upstream neighbors, the FKBP domain was fused in place of the stop codon at the C terminus of each protein. Recombinant BACs were transfected into two cultures that were grown in the presence and absence of Shield-1. The UL48 recombinant (UL48-FKBP) produced many fewer and much smaller plaques in the absence of Shield-1 (data not shown), providing a preliminary indication that the FKBP domain was having the desired effect. As expected because UL103 deletion mutants are viable, the UL103 recombinants (UL103-FKBP and UL103-FKBP-V5) produced infectious virions under both conditions. We were not able to reconstitute a viable virus from a UL94-FKBP BAC.

FIG 4.

Gene arrangements and transcripts in the vicinity of the candidate regulators of cVAC biogenesis. The 5′ ends of mRNAs are indicated with circles and 3′ polyadenylated regions with An. Genetic-recombineering methods described in Materials and Methods were used to introduce the FKBP degradation domain into our genes of interest. All three genes overlap neighboring genes at their 5′ ends. To avoid effects on overlapping coding regions at the N termini of the three genes, the FKBP destabilization domain was added at the C terminus of each gene. UL48 maps (49), UL94 maps (50–52), PRV UL7 (HCMV UL103 homolog) map (49), and HCMV UL103 map (53).

We used immunofluorescence and immunoblotting to verify our ability to regulate the stability of the proteins of interest. pUL48 was detected using an antibody against its N terminus (kindly provided by Wade Gibson) (31). Because highly specific antibodies are not available for detection of pUL103, a 14-amino-acid V5 epitope tag was added to the C terminus of UL103-FKBP (UL103-FKBP-V5). For UL48-FKBP, addition of Shield-1 markedly increased the amount of pUL48 detected by immunofluorescence (Fig. 5A) (6.2-fold per cell based on 3D quantification of at least 8 cells for each condition). By densitometry, there was a 3.1-fold increase of pUL48 in the presence of Shield-1 (Fig. 5C). The amount of pUL48 expressed by the UL48-FKBP virus in the presence of Shield-1 is less than that produced by the parental virus. When UL103-FKBP-V5 is grown in the absence of Shield-1, the UL103-FKBP-V5 protein is very faint and dispersed when visualized by immunofluorescence, with marked increases when Shield-1 is present (3.1-fold by IFA and 4.2-fold in immunoblots) (Fig. 5B and D). Thus, addition of the FKBP destabilization domain to pUL48 and pUL103 enables regulation of their stability. Immunoblots showed little difference in expression of virus genes from all three lytic classes in the presence or absence of pUL48 or pUL103 stabilization (Fig. 6), and only modest differences were seen in intracellular DNA levels by quantitative PCR (data not shown).

FIG 5.

Verification of regulated protein degradation. The UL48-FKBP- and UL103-FKBP-V5-tagged proteins were analyzed by confocal immunofluorescence microscopy (A and B) and immunoblotting in the presence and absence of Shield-1 (C and D). HFFs were infected for 120 h at an MOI of 0.1. pUL48 was detected using an antibody against its N terminus (C), and pUL103 was detected using an anti-V5 antibody (D).

FIG 6.

Effect of protein destabilization on other HCMV proteins. Cells were infected (MOI = 0.1) with UL48-FKBP and UL103-FKBP in the presence and absence of Shield-1 and then analyzed by immunoblotting at 24 and 120 h p.i., using antibodies against representative immediate-early (IE1 and IE2), early (UL44), and late (pp28/UL99) viral proteins. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was used as a loading control.

Having verified the intended effects of Shield-1 on viral proteins, the effect of regulated protein degradation on cVAC structure was analyzed by staining viral and cellular markers of the cVAC in the presence or absence of Shield-1 (Fig. 7 and 8). Shield-1 had no significant effect on cVAC development for the parental virus (Fig. 8). pUL48 destabilization in the absence of Shield-1 led to dispersal of early/recycling endosomes and absence of the characteristic Golgi ring, as was seen in the siRNA experiments. cVAC biogenesis appeared to be normal in most cells in which pUL48 was stabilized with Shield-1. Similar results were seen for UL103-FKBP. In an analysis of over 800 cells, cVAC disruption in the absence of Shield-1 was statistically significant (P < 0.0001) (Fig. 8). Thus, in accordance with the siRNA screen, destabilization of pUL48 and pUL103 disrupted cVAC formation, verifying their importance in cVAC biogenesis.

FIG 7.

Effects of regulated protein degradation on cVAC biogenesis. Cells were infected with the indicated viruses in the presence or absence of Shield-1 for 120 h and then stained for cellular markers of cVAC (cytoplasmic staining) and infection (IE2; nuclear staining). Shield-1 had no effect on assembly complex development for the parental virus, whereas destabilization of pUL48 and pUL103 led to dispersal of early/recycling endosomes and absence of the characteristic Golgi ring.

FIG 8.

Relative abundances of regular versus irregular cVAC structures in the presence and absence of Shield-1. Regular structures displayed circular Golgi rings and a concentration of EEA1-positive vesicles inside the rings. Irregular structures had fragmented or abnormal Golgi shapes and/or dispersal of EEA1-positive vesicles. Statistical significance was measured using the chi-square test (***, P < 0.0001). n, number of cells counted.

Growth properties of recombinant viruses in the presence and absence of Shield-1.

The siRNA and protein stability experiments described above were done at low multiplicity and employed cVAC morphology endpoints visible in infected cells at 4 or 5 days p.i. To determine whether destabilization of pUL48 or pUL103 influences virus growth, we compared the growth of three recombinant viruses (UL48-FKBP, UL103-FKBP, and UL103-FKBP-V5) with that of their parent (pAD/Cre). Time courses of production of extracellular infectious virions were examined at low and high MOI (multi- and single-step growth curves, respectively). At low MOI, not all cells are infected at the time of inoculation. Thus, multiple rounds of replication can take place, enabling examination of virus dissemination in terms of production of secondary plaques, features such as plaque size, and dependence on conditions created when cells are not exposed to large numbers of viral particles, many of which are not independently infectious (e.g., HCMV dense bodies and noninfectious enveloped particles). At high MOI, essentially every cell in the culture is infected simultaneously, and cells are also exposed to large numbers of bioactive noninfectious particles. Such synchronized infections provide information about how much infectious virus is produced during a single round of replication.

For UL48-FKBP, the single- and multistep growth infections were done in the presence or absence of Shield-1 (Fig. 9A). The titer of progeny extracellular virions was determined in triplicate in the presence of the drug. Somewhat unexpectedly, the single-step growth curves for UL48-FKBP showed 1-log-unit growth defects starting at 3 days p.i., regardless of whether Shield-1 was present or absent; the defect persisted over the 6-day experiment. Likewise, in the multistep analysis, UL48-FKBP growth rates were similar in the presence and absence of Shield-1; mutant infectious progeny were first detected at 6 days p.i., and infectious yields lagged by ∼3 log units until 15 days p.i., when they were about 1 log unit lower than for the parental virus. Inefficient growth of the UL48-FKBP virus might be attributable in part to the lower overall level of pUL48 expressed in the presence or absence of Shield-1 (Fig. 5C).

FIG 9.

Growth properties of recombinant extracellular viruses in the presence and absence of Shield-1. (A) Replication of UL48-FKBP. (B) Replication of UL103-FKBP and UL103-FKBP-V5. Shown is virus replication at high (MOI = 3) and low (MOI = 0.01) MOI (single step and multistep growth curves, respectively) in the presence or absence of Shield-1. +, cells infected in the presence of Shield-1; −, cells infected in the absence of Shield-1. Virus titers were determined in triplicate; the standard errors are too small to be seen in the graphs.

The UL103 recombinant virus stocks were also grown with Shield-1. In single- and multistep analyses, the presence of Shield-1 had little effect on the yield of extracellular infectious virus (Fig. 9B). Titers were similar when measured in the presence or absence of Shield-1 (data not shown), the only difference being plaque sizes (see below). In the single-step experiment in the absence of Shield-1, the UL103-FKBP recombinant virus produced infectious extracellular virus at rates similar to those of its parent at most time points, with a <10-fold difference at 3 days p.i. In contrast, Ahlqvist and Mocarski found up to a 50-fold decrease in extracellular infectious titer with a Towne-BAC mutant that does not express pUL103 (15). These results suggest that at high multiplicities of infection, there is some compensation for the absence of pUL103 by a complementary function likely provided by another viral protein. In multistep experiments, the UL103-FKBP and UL103-FKBP-V5 viruses lagged in extracellular virus production by 2 to 3 log units at 3 and 6 days p.i., ultimately reaching parental titers 9 to 15 days p.i. In marked contrast, a pUL103-null virus produced an ∼3,000-fold lower extracellular infectious titer (15). This indicates that even small amounts of pUL103 expressed by the UL103-FKBP virus are sufficient to support nearly parental growth. The latter point is supported by our observations that (i) at low multiplicities, the small amount of pUL103 produced by our FKBP-tagged viruses is sufficient to rescue about 2 log units of the early (3- to 6-day-p.i.) growth defect seen in the absence of pUL103, with the growth defect being overcome 9 to 15 days p.i., and (ii) the growth of these viruses at an MOI of 0.1 was similar to that in the single-step analysis (data not shown); the growth defect was seen only at low MOI.

In other growth experiments with FKBP-tagged UL48 and UL103 viruses, we saw little effect of Shield-1 over concentrations ranging from 0 to 4 μM (data not shown). These results suggest that although the FKBP domain enables regulated stability of pUL48 and pUL103, as well as regulation of their activities in cVAC biogenesis, for both proteins, the FKBP tag inhibits some aspect of production or egress of infectious virions. As mentioned above, for pUL103, this effect was largely overcome by 9 to 15 days p.i.

Ahlqvist and Mocarski found that viruses deficient in expression of pUL103 produce significantly smaller plaques (15). Consistent with this, we found that Shield-1 could rescue the small-plaque phenotype of the UL103-FKBP virus (Fig. 10). Thus, in addition to its roles in cVAC biogenesis and virion maturation or egress, pUL103 has a role in cell-to-cell spread of infection. In contrast, the UL48 recombinant virus produced small plaques in the presence and absence of Shield-1 (data not shown).

FIG 10.

Plaque sizes of UL103 recombinant viruses in the presence and absence of Shield-1 at 10 days p.i. The plaque size (area) (± standard error of the mean [SEM]) was measured from photographs by using ImageJ (NIH) (54). In the absence of Shield-1, UL103-FKBP plaque sizes were significantly smaller than for the other three conditions (t test; P < 10⁻⁵).

In ultrastructural analyses, in the absence of Shield-1, cytoplasmic UL103-FKBP virions were more likely to have aberrant appearances, including frequent appearance of virions that seemed to be arrested or greatly slowed down in the midst of acquiring their envelopes (Fig. 11).

FIG 11.

(A) Comparison of regular versus irregular virions inside infected cells in the presence or absence of Shield-1. (B) Regular virions contain DNA and an envelope. Irregular virions are in the process of envelopment or display an abnormal structure. Statistical significance was measured using the chi-square test (*, P < 0.1; ***, P < 0.0015). n, number of viral particles counted.

DISCUSSION

Little is known about the effectors of HCMV cVAC biogenesis, a biological process that is distinct from cVAC operation. Structures architecturally similar to the cVAC (Golgi rings surrounding a cluster of early/recycling endosomes and spatial segregation of the early/recycling and late endosomal systems) have been identified in uninfected cells (32–34). cVAC biogenesis may thus be a product of the virus activating, and possibly modifying, a previously programmed biological process. Like other biological pathways regulated by HCMV (e.g., immune responses, apoptosis, and the AKT/PI3K/mTOR pathway) (35, 36), we anticipate multiple points of control of the process by the virus, including the recently described role of HCMV microRNAs in cVAC biogenesis (25). In addition to the virologic lessons, elucidation of this pathway and its points of control will be informative with respect to important areas of cell biology.

Using siRNAs, we screened HCMV early-late and late genes known to be important for efficient lytic replication for their activities in cVAC biogenesis. Of the 26 genes tested, only inhibition of expression of UL48, UL94, and UL103 specifically prevented cVAC biogenesis: Golgi rings did not form and the early/recycling endosome complex was disrupted in the context of infections that had proceeded to at least late-gene expression.

The genes for all three candidate cVAC regulators have coding regions that overlap coding regions of their neighboring genes, and polycistronic transcripts span each gene locus (Fig. 4). To address the possibility that the siRNA effects were due to silencing of overlapping transcripts, we tagged each candidate protein with a regulatable posttranslational destabilization domain. Although we were not able to reconstitute a viable virus from a BAC containing a tagged version of UL94, viable recombinants were reconstituted for UL48 and UL103, and we demonstrated regulated stability of the proteins of interest. HCMV immediate-early (IE), early, and late genes were expressed at near-parental levels regardless of pUL48 or pUL103 degradation. When pUL48 or pUL103 was destabilized, cVAC biogenesis was impaired in a manner similar to what was seen following siRNA treatment, thus confirming the siRNA result and verifying its specificity for these genes.

Our siRNA results are consistent with prior reports that HCMV UL94 and UL103 are important for late stages of HCMV virion assembly and egress (15, 37). From the UL103-FKBP virus, we learned that the pUL103 domain necessary for cVAC biogenesis is distinct from at least two other functional domains. Specifically, defects in plaque size (Fig. 10) and secondary envelopment (Fig. 11) were able to be rescued by Shield-1, and the C-terminal FKBP tag appears to alter production of extracellular infectious virus in a manner that is impervious to, and possibly worsened by, stabilization of the tagged protein with Shield-1 (Fig. 9B). We also provide new evidence that HCMV pUL48 is important for production of infectious extracellular virus, perhaps analogous to the role of its herpes simplex virus 1 homolog in production of noninfectious light bodies (38). Similar to the behavior of the UL103-FKBP virus, this defect in the UL48-FKBP virus was not rescued by Shield-1, in contrast to the cVAC biogenesis activity. More detailed studies of the roles of these proteins in cVAC biogenesis will require generation of reagents that do not include tags that interfere with functions of the proteins of interest.

The three HCMV proteins whose silencing prevented cVAC formation are virion tegument proteins that are conserved across the Herpesviridae (Table 6). It has been suggested that HCMV UL103 and its orthologs in other herpesviruses play related roles in the final stages of virion assembly (15). While structures analogous to the HCMV cVAC have not been identified for other herpesviruses, is possible that these conserved proteins are all involved in remodeling infected cells into virus production factories.

It is clear that multiple viral and cellular gene products play important roles in cVAC biogenesis. We hypothesize that collective interactions of these factors, with each other and with other viral and cellular proteins, establish an environment conducive to cVAC development. The viral proteins identified here as regulators of cVAC biogenesis will be useful probes for identifying and understanding the network of viral and cellular genes involved in cVAC biogenesis and virion maturation, a necessary prelude to developing novel antivirals that block the process.