Antagonistic Determinants Controlling Replicative and Latent States of Human Cytomegalovirus Infection

ABSTRACT

The mechanisms by which viruses persist and particularly those by which viruses actively contribute to their own latency have been elusive. Here we report the existence of opposing functions encoded by genes within a polycistronic locus of the human cytomegalovirus (HCMV) genome that regulate cell type-dependent viral fates: replication and latency. The locus, referred to as the UL133-UL138 (UL133/8) locus, encodes four proteins, pUL133, pUL135, pUL136, and pUL138. As part of the ULb′ region of the genome, the UL133/8 locus is lost upon serial passage of clinical strains of HCMV in cultured fibroblasts and is therefore considered dispensable for replication in this context. Strikingly, we could not reconstitute infection in permissive fibroblasts from bacterial artificial chromosome clones of the HCMV genome where UL135 alone was disrupted. The loss of UL135 resulted in complex phenotypes and could ultimately be overcome by infection at high multiplicities. The requirement for UL135 but not the entire locus led us to hypothesize that another gene in this locus suppressed virus replication in the absence of UL135. The defect associated with the loss of UL135 was largely rescued by the additional disruption of the UL138 latency determinant, indicating a requirement for UL135 for virus replication when UL138 is expressed. In the CD34⁺ hematopoietic progenitor model of latency, viruses lacking only UL135 were defective for viral genome amplification and reactivation. Taken together, these data indicate that UL135 and UL138 comprise a molecular switch whereby UL135 is required to overcome UL138-mediated suppression of virus replication to balance states of latency and reactivation.

IMPORTANCE Mechanisms by which viruses persist in their host remain one of the most poorly understood phenomena in virology. Herpesviruses, including HCMV, persist in an incurable, latent state that has profound implications for immunocompromised individuals, including transplant patients. Further, the latent coexistence of HCMV may increase the risk of age-related pathologies, including vascular disease. The key to controlling or eradicating HCMV lies in understanding the molecular basis for latency. In this work, we describe the complex interplay between two viral proteins, pUL135 and pUL138, which antagonize one another in infection to promote viral replication or latency, respectively. We previously described the role of pUL138 in suppressing virus replication for latency. Here we demonstrate a role of pUL135 in overcoming pUL138-mediated suppression for viral reactivation. From this work, we propose that pUL135 and pUL138 constitute a molecular switch balancing states of latency and reactivation.

INTRODUCTION

Human cytomegalovirus (HCMV) is a ubiquitous betaherpesvirus. Like all herpesviruses, HCMV persists indefinitely in the immunocompetent host by establishing latency in specialized cellular reservoirs. While primary infection with HCMV is typically asymptomatic, HCMV reactivation in immunocompromised individuals, including solid organ or stem cell transplant patients, is a significant cause of morbidity and mortality (1–3). Further, congenital infection occurs in 1 in 150 children in the United States, making HCMV the leading cause of infectious disease-related birth defects (4–7). The health cost of lifelong HCMV persistence is beginning to emerge as age-related pathologies associated with chronic inflammation, including atherosclerosis, immune senescence, and frailty (8–18). Understanding virus-host interactions important to HCMV persistence is essential to ultimately controlling the virus.

Clinical strains of HCMV contain a 13- to 15-kb region of the genome, termed ULb′, that is lost upon serial passage of the virus in cultured fibroblasts (19–21). Therefore, laboratory-adapted strains of HCMV lack the ULb′ region, and the 20 putative open reading frames (ORFs) encoded within ULb′ have been deemed nonessential for replication in fibroblasts (19, 20). Consequently, ULb′ genes have been understudied, and functions have been defined for very few. Because the ULb′ genes have been retained through the evolution of HCMV and are conserved in all clinical strains, these genes are hypothesized to play important roles in viral persistence in the host, functioning in immune evasion, replication, or latency in cell types other than fibroblasts. Consistent with this hypothesis, viruses lacking the ULb′ region are altered in their ability to cause disease (22), to establish/maintain latency (23, 24), and to infect clinically relevant cells, including endothelial cells, epithelial cells, and leukocytes (22, 25, 26).

The UL133-UL138 polycistronic locus (herein referred to as UL133/8 locus) is encoded by genes within the ULb′ region and is important for latency in the experimental CD34⁺ hematopoietic progenitor cell (HPC) model of latency (27). The UL133/8 locus encodes four novel proteins, pUL133, pUL135, pUL136, and pUL138 (28), each of which is associated with Golgi apparatus membranes by N-terminal transmembrane domains that result in the large C-terminal domain being oriented on the cytosolic side of Golgi apparatus membranes (27, 29). pUL138 promotes a latent infection in primary CD34⁺ HPCs infected in vitro (23, 29). pUL138 physically interacts with both pUL133 and pUL136 (30). The pUL133-pUL138 complex appears to cooperatively function in promoting a latent infection, as viruses containing disruptions in pUL133, pUL138, or both replicate with increased efficiency in CD34⁺ cells (27, 30). pUL138 has been shown to increase cell surface levels of TNFR (31, 32) and decrease surface levels of MRP-1 (33), although the significance of these surface alterations to viral infection is not completely understood. The roles of pUL135 and pUL136 have not yet been described.

In the present study, we describe the existence of a novel molecular switch comprised of UL135 and UL138 that balances states of latency and viral replication. We demonstrate a profound requirement for UL135 for reconstitution of virus replication from infectious bacterial artificial chromosome (BAC) clones of the HCMV genome in fibroblasts when UL138 is expressed. While the requirement for UL135 for replication can be overcome in fibroblasts at high multiplicities of infection (MOIs), UL135 is required for viral genome amplification and virus replication and reactivation in CD34⁺ HPCs. The phenotypes associated with the UL135-deficient virus were at least partially overcome by the additional disruption of UL138. Our data demonstrate a complex interplay between pUL135 and pUL138 in regulating virus replication. The study represents the first demonstration of a requirement for a ULb′ gene in virus replication and defines antagonistic viral factors balancing states of latency and reactivation.

MATERIALS AND METHODS

Cells.

Primary human embryonic lung fibroblasts (MRC-5 cells purchased from ATCC, Manassas, VA) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 8% fetal bovine serum, 10 mM HEPES, 1 mM sodium pyruvate, 2 mM l-alanyl-glutamine, 0.1 mM nonessential amino acids, 100 U/ml penicillin, and 100 μg/ml streptomycin. Human cord blood was obtained from donors at the University Medical Center at the University of Arizona using a protocol approved by the Institutional Review Board. These specimens were completely deidentified and provided to our research group as anonymous samples. Mononuclear cells and CD34⁺ HPCs were isolated and cultured as previously described (28, 29). CD34⁺ cells were maintained in long-term culture as described previously (29) but using MyeloCult H5100 medium (Stem Cell Technologies). The M2-10B4 murine stromal cell line expressing human interleukin-3 (IL-3) and granulocyte colony-stimulating factor (G-CSF) and the Sl/Sl murine stromal cell line expressing human IL-3 and stem cell factor (SCF) (a kind gift from Stem Cell Technologies on behalf of D. Hogge, Terry Fox Laboratory, University of British Columbia, Vancouver, BC, Canada) were cultured as recommended previously (34). All cells were maintained at 37°C with 5% CO₂.

Viruses.

BAC clones of the HCMV TB40/E (35) or FIX (VR1814) (36) parental (wild-type [WT]) viruses were engineered to express the green fluorescent protein (GFP) as a marker of infection (27). The construction of a mutant unable to express UL138 (UL138_STOP) and a mutant with a deletion of the entire UL133/8 locus (UL133-UL138_NULL [herein referred to as UL133/8_NULL]) was described previously (27, 29). The oligonucleotide primers used in the construction of recombinant viruses are listed in Table 1. BAC genomes were maintained in Escherichia coli, and recombinant viruses were generated through a two-step, positive-negative selection approach that leaves no trace of the recombination process (29, 37, 38). To create the kan/lacZ or galK substitutions in the viral BAC genome, the kan/lacZ or galK cassette was amplified by PCR using primers flanked by homologous viral sequences and recombined into the viral BACs as described previously (27–29).

TABLE 1.

Primers used in this work

^aFwd, forward; Rev, reverse. Asterisks in the primer name indicate conversion of the indicated amino acid codon to a stop codon.

^bHCMV sequences are in uppercase, and kan/lacZ or galK sequences are in lowercase; [PHOS], 5′ phosphorylated; enzyme sites are shaded gray; stop codons that were replaced are in bold; mutated methionines are italicized.

To introduce stop codons into the viral BAC genome, we created a shuttle vector to capture sequences from UL133 through UL138 from TB40/E. A segment from UL148A to UL133 was amplified from TB40/E BAC with UL148A_Fwd+16/EcoRV and UL133_Rev-239/HindIII (Table 1). A segment from UL138 to the 3′ untranslated region of UL138 was amplified with UL138_Fwd/HindIII and UL138_Rev+535/EcoRV. PCR products of UL148a-UL133 and UL138 were combined as the template for an overlap extension PCR, using primers UL148A_Fwd+16/EcoRV and UL138_Rev+535/EcoRV. The product was gel purified, A tailed with Taq, and TA cloned into the pGEM-T Easy vector (Promega, Madison, WI) to yield pGEM-T-ULb′-empty. The plasmid pGEM-T-ULb′-empty was linearized with HindIII and electroporated into bacteriophage lambda Red-induced E. coli SW102 harboring the TB40/E BAC genome (chloramphenicol resistant) to retrieve the ULb′ region from UL148A to UL138 by allelic exchange, yielding pGEM-T-UL133/UL138 (ampicillin resistant). Plasmid pGEM-T-UL133/UL138 was confirmed by sequencing the complete UL133 to UL138 insert. ATG codons were mutated to TAG/stop codons by site-directed Phusion mutagenesis as recommended by the manufacturer (NEB), and mutations were confirmed by sequencing the UL133-UL138 insert. The following methionine residues were mutated: in UL133, M1STOP and M57STOP; in UL135, M1STOP, M21STOP, and M97STOP of the 328-amino-acid annotation available through NCBI (GenBank accession number ABV71656); and in UL138, M1STOP. These mutated sequences were used to generate the HCMV TB40/E BACs described in Fig. 1. Mutated versions of the UL133-UL138 fragment were released from pGEM-T-UL133/UL138 with EcoRV, gel purified, and recombined into UL133/8_NULL as previously described (29). BAC integrity was verified by restriction fragment analysis, and the UL133-UL138 region was sequenced. TB40/E-UL135myc was created by cloning the myc epitope tag in frame onto the C terminus of pUL135 in TB40/E with the primers and methods described previously (28).

FIG 1.

Construction of FIX and TB40/E recombinant viruses and the requirement for UL135 for virus replication. (A and B) Schematic of the ULb′ locus depicting UL150-UL138 locus genes in HCMV strains FIX (A) and TB40/E (B). Gray arrows, WT genes in the locus; black arrows, kan/lacZ or galK cassette replacing the indicated UL150-UL135 genes (the drawing is not to scale); white arrow, genes containing stop codon substitutions (denoted by asterisks) replacing the ATG codon(s); plus and minus symbols to the right, the replication phenotype of the recombinant viruses. In FIX(ur)-UL135_M1-STOP, the first ATG codon (Met1) was replaced by a stop codon. In TB40/E viruses, the initiating codon was replaced by a STOP codon in UL133_STOP, UL135_M1-STOP, and UL138_STOP viruses. TB40/E-UL135_STOP contains stop codons substituted for methionine codons at positions M1, M21, and M97. US and UL, unique short and unique long regions, respectively. (C) BAC genomes for TB40/E WT and/or the indicated mutant viruses expressing GFP as a marker for infection were transfected into fibroblasts by electroporation. Infected GFP⁺ cells and plaques are shown at 10 and 30 days posttransfection.

Virus stocks were generated by transfecting 15 μg of the viral BAC genome and 2 μg of a plasmid carrying UL82 (pp71) into 5 × 10⁶ MRC-5 fibroblasts and stored as described previously (29). Virus titers were determined by measurement of the 50% tissue culture infective dose (TCID₅₀) on MRC-5 fibroblasts.

Plasmids.

To analyze the methionine utilization within the UL135 ORF, the open reading frame was amplified from TB40/E using primers UL135-25nt_Fwd/XbaI and UL135_Myc Rev/BamHI and cloned into the pCIG expression vector. Methionine residue 1, 21, and/or 97 was mutated to leucine (ATG to CTG) by serial Phusion mutagenesis (primer sequences are indicated in Table 1). All sequences were confirmed by restriction fragment analysis and sequencing of the UL135 insert. MRC-5 cells (2 × 10⁶) were transfected with 2 μg of each plasmid by electroporation in a 2-mm cuvette at 130 V and a time constant of 30 ms. Two days after transfection, cells were trypsinized, washed twice with phosphate-buffered saline (PBS), and lysed in radioimmunoprecipitation assay (RIPA) buffer for Western blotting.

Immunoblotting.

Immunoblotting was performed as described previously (29). Briefly, 15 to 20 μg of protein lysates was separated on 12% or 4 to 20% bis-Tris polyacrylamide gels by electrophoresis in MES (morpholineethanesulfonic acid) buffer and transferred to 0.45-μm-pore-size polyvinylidene difluoride membranes (Immobilon-FL; Millipore, MA). These PAGE conditions differentially resolve the pUL135 bands seen in immunoblots. The proteins were immunoblotted using mouse monoclonal antibodies or rabbit polyclonal antibodies directed against each protein and detected using fluorescently conjugated secondary antibodies and an Odyssey infrared imaging system (LI-COR, Lincoln, NE). All antibodies used are listed in Table 2. A polyclonal antibody to pUL135 was generated by Open Biosystems, using a peptide corresponding to amino acids 297 to 313 of pUL135 (ELAPPPRWSDIEELLEK). Antibodies to pUL135 were affinity purified from serum using the immunizing peptide coupled to CarboxyLink resin (Pierce Biotechnology, IL), as described previously (28).

TABLE 2.

Primary antibodies used for immunoblotting

Antigen	Antibody (clone)	Typea	Source	Dilution for blotting
IE1/2	3H4	M	Giftb	1:100
pUL133	Custom	R	Open Biosystems	2 μg/ml
pUL135	Custom	R	Open Biosystems	2 μg/ml
pUL138	Custom	R	Open Biosystems	2 μg/ml
pUL44	10D8	M	Virusys CA006	1:2,500
pp28 (UL99)	10B4-29	M	Gift	1:50
α-Tubulin	DM1A	M	Sigma T9026	1:20,000
MCP (pUL86)	127	M	Gift	1:70
pp65 (pUL83)	8F5	M	Gift	1:25

^aR, rabbit polyclonal antibody; M, mouse monoclonal antibody.

^bGift, a generous gift from Tom Shenk, Princeton University.

Virion purification.

Cell-free TB40/E virus particles, noninfectious enveloped particles (NIEPs), and dense bodies (DBs) were purified through positive-density/negative-viscosity glycerol-potassium tartrate gradients (30% glycerol, 15% potassium tartrate to 40% potassium tartrate in TN buffer [150 mM NaCl and 10 mM Tris-Cl, pH 8.0]), as described previously (39). A total of 5 × 10⁷ PFU was loaded on top of the gradient in a Beckman SW40 rotor and spun at 36,000 rpm for 15 min at 20°C, before removing NIEPs. The virions were separated from DBs by five additional hours of spinning at 36,000 rpm. Each fraction of particles was washed twice with PBS and pelleted by spinning for 60 min at 36,000 rpm in the SW40 rotor and then lysed, boiled in reducing sample buffer, and separated on a 10% bis-Tris polyacrylamide gel.

Viral genome quantification.

For genome-to-PFU measurements, equal numbers of PFU of cell-free WT and mutant viruses were treated with 0.2 mg/ml proteinase K in lysis buffer (20 mM Tris-HCl, pH 8.0, 200 mM NaCl, 50 mM EDTA) for 3 h at 60°C. DNA was extracted with phenol-chloroform and ethanol precipitated. For quantitating viral genomes in infected cells, total genomic DNA was isolated by phenol-chloroform extraction or by use of a genomic DNA miniprep kit (Zymo Research). In both experiments, viral genomes were quantitated by quantitative PCR (qPCR) using the LightCycler 480 Probes master mix (Roche) according to the manufacturer's instructions, Universal ProbeLibrary (UPL; Roche) probe 55, and primers specific to the IE2 gene (primers IE2-Fwd [′-CCAGCGTCGCGGTTACTA-3′] and IE2-Rev [5′-TTTCTTACGCGGGCTGAG-3′]). To determine the number of genomes present in a given number of PFU, viral DNA copy numbers were estimated by an absolute quantification method using known IE2 DNA standards. Viral genomes synthesized during infection were determined using human GAPDH (glyceraldehyde-3-phosphate dehydrogenase) as a reference (primers GAPDH-Fwd [5′-TCTTAAAAAGTGCAGGGTCTGG-3′] and GAPDH-Rev [5′-AGAGTTGTCAGGGCCCTTTT-3′] with UPL probe 83).

Transmission electron microscopy (TEM).

MRC-5 cells were mock infected or infected at an MOI of 2. Cells were fixed in 2.5% glutaraldehyde and 0.1 M PIPES [piperazine-N,N′-bis(2-ethanesulfonic acid)] at 2 days postinfection. The fixed cell pellet was postfixed with osmium tetroxide in 0.1 M PIPES and dehydrated in a graded series of alcohol. Pellets were infiltrated with resin and cut into 100-nm sections. The sections were floated onto copper grids and imaged using a Phillips CM-12 transmission electron microscope. Cells were embedded and sectioned by the University Spectroscopy and Imaging Facility of Arizona Research Laboratories.

Infectious centers assay.

CD34⁺ HPCs isolated from human cord blood were infected at an MOI of 2 for 20 h in Iscove's modified Dulbecco's medium supplemented with 10% BIT9500 serum substitute (Stem Cell Technologies, Canada), 2 mM l-glutamine, 20 ng/ml low-density lipoproteins (Calbiochem), and 50 μM 2-mercaptoethanol. Following infection, pure populations of infected CD34⁺ HPCs (>98% GFP positive [GFP⁺]) were isolated by fluorescence-activated cell sorting (FACSAria; BD Biosciences Immunocytometry Systems, San Jose, CA) using a phycoerythrin-conjugated antibody specific to CD34 (BD Biosciences). Cells were sorted by the University of Arizona Shared Service at the University of Arizona Cancer Center. Pure populations of infected HPCs were cultured in transwells above an irradiated (3,000 rads; ¹³⁷Cs gammacell-40 irradiator type B; Best Theratronics, Ottawa, ON, Canada) M2-10B4 and Sl/Sl stromal cell monolayer (34) for 10 to 12 days. The frequency of infectious center production was measured using a limiting dilution assay as described previously (23). Briefly, infected HPCs were serially diluted 2-fold in long-term bone marrow culture (LTBMC) medium supplemented with 15 ng/ml each of interleukin-6, granulocyte colony-stimulating factor, and granulocyte-macrophage colony-stimulating factor (R&D Systems, MN). Aliquots of 0.05 ml of each dilution were added to 12 wells (the first dilution corresponds to 40,000 cells per well) of a 96-well tissue culture plate containing MRC-5 cells. To differentiate virus made as a result of reactivation from virus preexisting in the long-term cultures, an equivalent number of cells was mechanically disrupted and seeded into MRC-5 cocultures in parallel with the reactivation experiments. MRC-5 cells were monitored for GFP expression for a period of 14 days. The frequency of infectious centers formed was calculated on the basis of the number of GFP⁺ cells at each dilution using extreme limiting dilution analysis (ELDA) software (http://bioinf.wehi.edu.au/software/elda) (40).

RESULTS

UL135 is required to reconstitute infection from transfection of infectious genomes.

pUL135 is one of four proteins expressed from the UL133/8 locus encoded by genes within the ULb′ region of the HCMV genome. Like all ULb′ sequences, the UL133/8 locus is largely dispensable for virus replication in cultured fibroblasts (27). At low multiplicities of infection, viruses lacking the UL133/8 locus replicate with a modest defect, which is largely overcome at higher multiplicities of infection. Therefore, we were surprised to discover that recombinant genomes containing disruptions in only UL135 were severely defective for reconstituting infection. In total, seven independent recombinant BACs were constructed and analyzed (Fig. 1A and B): two BACs lacking 2 to 5 kb, including UL135, in the HCMV FIX strain [FIX(ur)-sub1 and FIX(ur)-sub1 short; Fig. 1A]; two viruses where the entire UL135 coding sequence was disrupted by kan/lacZ or galK substitution in strains FIX [FIX(ur)-UL135_NULL; Fig. 1A] and TB40/E (TB40/E-UL135_NULL; Fig. 1B), respectively; and three viruses containing stop codon substitutions for predicted 5′ methionine codons for UL135 in strains FIX [FIX(ur)-UL135_M1-STOP; Fig. 1A] and TB40/E (TB40/E-UL135_M1-STOP and -UL135_STOP; Fig. 1B). With the exception of FIX(ur)-UL135_M1-STOP and TB40/E-UL135_M1-STOP, each of these viruses exhibited a replication defect such that infectious virus could not be reconstituted from the transfection of infectious BAC clones of the viral genomes. Transfection of the BAC genomes containing disruption of UL135 resulted in the production of single GFP-positive cells or small plaques (2 to 5 cells), but the virus did not spread over the course of 30 days (data not shown and Fig. 1C). These data are representative of the data acquired from five independent transfections of six independent UL135_NULL or UL135_STOP BAC clones in either the TB40/E or the FIX strain (data not shown). This phenotype is unique to UL135, as other ORFs encoded within the UL133/8 locus are not required for replication in fibroblasts (27).

UL135 is expressed as multiple isoforms that are not included in the virus particle.

The UL135 protein, pUL135, migrates as three distinct bands in cells infected with WT TB40/E or with TB40/E encoding a C-terminal myc epitope-tagged version of pUL135 (Fig. 2A, lane 1). pUL135 was detected either with a polyclonal antibody to a pUL135 peptide that we generated or with a monoclonal antibody to the myc epitope tag cloned in frame with the C terminus of pUL135. The third isoform (which had the lowest molecular mass) was inconsistently detected from experiment to experiment and was detected at later time points of infection relative to the times of detection of the larger isoforms (data not shown). Two UL135 annotations exist predicting either a 328-amino-acid protein (NCBI GenBank accession number ABV71656) or 308-amino-acid protein (NCBI GenBank accession number ADE88119), depending on the initiation at methionine codon 1 (M1) or M21 of the 328-amino-acid annotation, respectively. The ability of FIX(ur)-UL135_M1-STOP and TB40/E-UL135_M1-STOP to replicate when other UL135 mutant viruses do could be explained if pUL135 initiates translation at a methionine codon downstream of the start codon predicted for the 328-amino-acid annotation.

FIG 2.

pUL135 isoforms are not incorporated into the virus particle. (A) UL135 annotations predict a 308- or 328-amino-acid protein, depending on the 5′ initiation codon. The 328-amino-acid sequence coding for pUL135 was cloned into an expression construct with a C-terminal myc epitope tag. ATG codons corresponding to M1, M21, or M97 of the 328-amino-acid annotation were mutated by site-directed mutagenesis. Expression of the three pUL135 isoforms (marked 1, 2, and 3) was detected by immunoblotting lysates from fibroblasts transiently expressing each construct (lanes 2 to 6). pUL135 was detected using an antibody specific to the myc epitope tag. Lysates derived from fibroblasts infected with TB40/E UL135myc served as a positive control for the pUL135 isoforms (lane 1). The infection lysate (lane 1) was run on the same gel as lysates in lanes 2 to 6 but was scanned at a lower intensity. Asterisks, background bands that were present in the negative control (lane 2). (B) Fibroblasts were mock infected or infected (MOI = 1) with FIX or TB40/E viruses containing wild-type UL135 or recombinant viruses containing stop codon substitutions for ATG codons corresponding to M1 (M1-STOP) or M1, M21, and M97 (STOP). Lysates harvested at 48 h postinfection were analyzed for UL135 expression by immunoblotting using rabbit polyclonal antiserum specific to pUL135. α-Tubulin served as a loading control. (C) WT virions were banded on glycerol-tartrate gradients, lysed, and analyzed for the inclusion of pUL135 or another UL133/8 protein, pUL133, by immunoblotting using polyclonal antiserum to pUL135 or pUL133. As controls for virion purity, the major capsid protein (MCP; pUL86) and pp65 tegument protein (pUL83), but not α-tubulin, were detected in virion preparations. An infected cell lysate (24 hpi) served as a positive control for detection of each protein by the antibodies used.

To map the translational start sites of pUL135, we cloned the entire UL135 coding sequence of the 328-amino-acid annotation into an expression construct. We then converted the ATGs coding for M1, M21, or M97 to CTG (leucine) and transiently expressed each construct in fibroblasts. The UL135-specific bands produced by the expression construct correspond to the UL135 bands that are produced during infection, but their relative ratios differ (Fig. 2A, lane 3). Mutation of the ATG coding for M21 and M97, leaving only the ATG for M1, ablated UL135 expression (Fig. 2A, lane 4). This result indicates that the codon for M1 is not used for initiation of translation. This result is consistent with that of our previous work mapping the UL135 transcript, which originates at the codon for M21 (28). Mutation of the ATG coding for M1 (Fig. 2A, lane 5) resulted predominantly in expression of isoform 2 and a minor amount of isoform 1 (Fig. 2A, lane 3), suggesting that this codon is not used for initiation of translation under these conditions. However, mutation of the ATG coding for M1 and M21 resulted in the loss of both higher-molecular-mass isoforms (isoforms 1 and 2) and the prominent expression of the lowest-molecular-mass isoform (Fig. 2A, lane 6). This result suggests that the ATG coding for M97, in addition to M21, can secondarily be used for translation of pUL135. Further, the highest-molecular-mass isoform (isoform 1) may represent a posttranslational modification of the isoform originating at M21, since isoforms 1 and 2 are both ablated by disruption of the codon corresponding to M21. Our results indicate that the 308-amino-acid annotation is correct and that the codon corresponding to M1 of the 328-amino-acid annotation is not used to initiate translation in this context.

To ensure that the ATG coding for M1 was not used in the context of virus infection, we compared UL135 expression in viruses where M1 or M1 in addition to M21 and M97 had been converted to stop codons to abrogate protein translation. Interestingly, mutation of M1 had no effect on pUL135 expression (Fig. 2B, lanes 3 and 6). Expression of all three pUL135 isoforms was abrogated when M21 and M97 were disrupted in addition to M1 (Fig. 2B, lanes 4 and 7). These data are consistent with those from transfection and indicate that the M1 predicted start codon of the 328-amino-acid annotation is not used during lytic infection of fibroblasts, further suggesting that the annotation of the 308-amino-acid protein is correct. This annotation is consistent with the previous data from our laboratory mapping the 5′ end of transcripts encoding UL135 (28). We cannot exclude the possibility that M1 is used in other contexts of infection. To ensure that we abrogated all possibilities for UL135 expression in subsequent experiments, we used the TB40/E-UL135_STOP virus containing stop codon substitution for ATGs encoding M1, M21, and M97. Further work is required to understand how each of these isoforms may independently contribute to infection.

We have reported the association of pUL135 with microsomal membranes in infected cells (27). Many viral tegument proteins are associated with secretory membranes and are incorporated into mature virions. The defect in reconstituting UL135_STOP virus from BAC transfection could be explained if pUL135 was incorporated into the virion and promoted virus replication following virus entry and prior to the onset of immediate early (IE) gene expression similarly to pp71 (41). To address this possibility, we analyzed glycerol-tartrate gradient-purified WT virions, noninfectious enveloped particles (NIEPs), and dense bodies (DBs) from cell-free tissue culture supernatant for the presence of pUL135 (Fig. 2C). NIEPs resemble virions, except that they lack viral genomes, reducing their density on the gradient relative to that of virions. DBs are infection-induced vesicles containing tegument proteins that have a greater density than virions. As a control for virion purity, we analyzed gradient bands corresponding to NIEPs, virions, and DBs for the presence of the major capsid protein (MCP) encoded by UL86, the pp65 tegument protein encoded by UL83, and the cellular protein α-tubulin. As anticipated, MCP and pp65 were abundant in gradient bands corresponding to NIEPs and virions. However, pp65, but not MCP, was abundantly detected in gradient bands corresponding to DBs. α-Tubulin was not detected in association with any of the three gradient bands, indicating the high purity of our fractions. We did not detect pUL135 or pUL133 in NIEP, virion, or DB bands using polyclonal antiserum specific to each protein, suggesting that these proteins are not packaged within the virion, nor are they contained in DBs. These results are consistent with our previous finding that pUL138 is not a constituent of purified virions (29). As these proteins are not constituents of the virion, de novo synthesis is required for their effect in infection.

pUL135 opposes UL138-mediated suppression of viral replication.

The finding that UL135 mutant viruses are severely defective for replication, whereas a virus lacking the entire locus, including UL135, replicates like wild-type strains is paradoxical. We hypothesized that at least one of the other UL133/8 proteins strongly suppresses viral replication in the absence of UL135. To test this hypothesis, we substituted stop codons for 5′ methionine codons in UL133, UL136, or UL138 in the TB40/E-UL135_STOP BAC genome (Fig. 1B). The resulting recombinant BAC genomes, TB40/E-UL133_STOP/UL135_STOP (TB40/E-3/5_STOP), TB40/E-UL135_STOP/UL136_NULL (TB40/E-5/6_NULL), and TB40/E-UL135_STOP/UL138_STOP (TB40/E-5/8_STOP), were analyzed for their ability to support productive virus replication following electroporation into MRC-5 fibroblasts. Similarly to TB40/E-UL135_STOP, TB40/E-3/5_STOP and TB40/E-5/6_NULL failed to replicate (Fig. 1C). While single GFP-positive cells were detected 10 days after electroporation of these BACs into fibroblasts, the infection had not progressed by 30 days posttransfection. In contrast, TB40/E-5/8_STOP BAC transfection resulted in virus replication, producing virus plaques similar to those produced by the wild-type virus by 30 days.

We further characterized the rescue of UL135_STOP replication achieved through the additional disruption of UL138 by performing multistep growth curve analysis. TB40/E-5/8_STOP replicated, but with a 5- to 10-fold defect relative to replication of the WT virus (Fig. 3A), indicating that disruption of UL138 partially restores the defect created by the disruption of UL135. Analysis of viral protein expression in cells infected with TB40/E-5/8_STOP demonstrated that neither pUL135 nor pUL138 was expressed (Fig. 3B). The 72-kDa and 86-kDa IE proteins, the pUL44 early protein, and the pp28 late protein were expressed comparably in TB40/E WT and TB40/E-5/8_STOP infections. To confirm the role for UL138 in suppressing virus replication in the absence of UL135, we reverted UL138 back to the wild-type sequence in TB40/E-5/8_STOP (TB40/E-5_STOP/8_REV). As expected, the replication defect associated with UL135_STOP was restored when UL138 was reverted back to the WT (Fig. 3C). Taken together, these results indicate that the defect in UL135_STOP replication can be largely overcome by disruption of UL138. Further, the requirement for UL135 for replication in fibroblasts is conditional; it is observed only in the context of UL138 expression.

FIG 3.

UL138 suppresses viral replication in fibroblasts in the absence of UL135. (A) Virus yields in lysates were measured by determination of the TCID₅₀s over a time course following MRC-5 infection with TB40/E WT and TB40/E-5/8_STOP viruses at an MOI of 0.02. Data points represent the averages of triplicate experiments, and standard deviations are shown. (B) Proteins expressed in MRC-5 cells infected with TB40/E WT and TB40/E-UL135_STOP/UL138_STOP viruses at an MOI of 1.0 were analyzed by immunoblotting to detect immediate early, early, and late proteins using monoclonal or polyclonal antibodies, as described in Table 2. (C) TB40/E-UL135_STOP/UL138_STOP and TB40/E-5_STOP/8_REV BAC genomes were transfected into MRC-5 cells to analyze viral replication and spread. Viral plaques (GFP⁺) are shown at 10 and 30 dpi.

A UL135-deficient virus is necessary to define the role of UL135 in infection. Generating a recombinant virus lacking UL135 is complicated because overexpression of UL135 outside the context of infection from transient transfection of an expression plasmid or from transduction of lentivirus expression vectors resulted in programmed cell death in fibroblasts (data not shown). Cells maintained in 4% serum, however, supported the reconstitution of infectious virus from transfection of UL135_STOP BAC DNA, and this reconstitution was decreased with increasing amounts of serum. The virus stocks used in all subsequent experiments were derived from the transfection of BAC and propagation of infectious virus in low levels of serum. The basis of the complementation of UL135_STOP replication in low levels of serum is not yet understood. We took several approaches to complement UL135_STOP by the conditional expression of UL135 (Cre dependent) only once cells became infected following transfection of UL135_STOP BAC engineered to express Cre recombinase (data not shown). However, these approaches were no more successful than propagating UL135_STOP under low-serum conditions.

Characterization of the UL135_STOP defect in replication.

The defect in UL135_STOP replication is reflected in an increased genome-to-PFU ratio relative to that for the WT virus (Fig. 4A). Fifteenfold more genome-containing particles of UL135_STOP were required to form a PFU compared to that required for the WT virus. In contrast, the UL138_STOP genome-to-PFU ratio was comparable to that for the WT. Consistent with the replication of 5/8_STOP, the genome-to-PFU ratio of UL135_STOP was partially restored with the additional disruption of UL138. The increased genome-to-PFU ratio of UL135_STOP indicates a possible defect in maturation, where defective particles accumulate.

FIG 4.

UL135_STOP virus replication. (A) Genome-to-PFU ratios were determined by quantifying the viral genomes isolated from equal numbers of PFU of cell-free TB40/E WT, TB40/E-UL135_STOP, TB40/E-UL138_STOP, and TB40/E-UL135_STOP/UL138_STOP viruses by qPCR. Bars represent the averages of four independent experiments, and standard deviations are shown. Student's t test was used to determine the significance of the values for each mutant infection relative to those for the WT (**, significance for UL135_STOP [P < 0.0001]; *, significance for 5/8_STOP [P < 0.02] and UL138_STOP [P > 0.5]). (B and C) The kinetics and yields of TB40/E-UL135_STOP replication relative to those for WT replication were determined by following infection of MRC-5 cells with an equal number of genomes (100 genomes per cell) (B) or an equal number of PFU (0.05 PFU per cell) (C). Cells and media were collected over a time course, and virus titers were measured by determination of the TCID₅₀s. Values represent the averages of three independent experiments, and standard deviations are shown. (D and F) Representative IE, early, and late proteins were analyzed over a time course of infection in MRC-5 cells infected with TB40/E WT or TB40/E-UL135_STOP virus at an MOI of 0.05 PFU per cell (D) or 1 PFU per cell (F). Proteins were detected by immunoblotting using the mouse or rabbit antibodies described in Table 2. (E) Viral yields were measured in lysates by determination of the TCID₅₀s at 0 (input), 3, and 6 days following infection of MRC-5 cells with TB40/E WT or TB40/E-UL135_STOP at a high MOI (1 PFU per cell). Bars represent the averages of three independent experiments, and standard deviations are shown.

When fibroblasts were infected with an equal number of genome-containing particles, the titer of input UL135_STOP was reduced 100-fold relative to that for WT (Fig. 4B), consistent with the altered genome-to-PFU ratio. Further, UL135_STOP exhibited a delay in replication relative to the time of replication of WT virus. At later points in the time course, WT infection, but not UL135_STOP infection, had produced input titers of virus (those at the zero time point), indicating a less productive UL135_STOP infection relative to WT infection. However, when infection was normalized to equal numbers of PFU (0.05 PFU per cell), UL135_STOP replicated comparably to WT virus in a multistep analysis (Fig. 4C). These results were surprising, given the difficulty of producing UL135_STOP virus stocks from transfection of infectious genomes. These results indicate that the UL135_STOP defect can be overcome by infection with an increased number of genome-containing particles or that components of the virion are important for rescuing the infectivity of UL135_STOP.

We further analyzed the accumulation of representative immediate early (IE1 and IE2), early (pUL44), and late (pp28) proteins, in addition to pUL135 and pUL138, over a time course following infection at 0.05 PFU per cell. As anticipated, pUL135 was not expressed in UL135_STOP infection. UL135_STOP-infected cells were delayed in the accumulation of the viral proteins analyzed, and peak levels did not match those for the WT for IE2 and pUL138 (Fig. 4D). This delay was realized as a slight reduction in virus yields between 5 and 8 days postinfection (dpi) (Fig. 4C). Interestingly, the ratio of IE1 to IE2 was much higher in UL135_STOP infection than in WT infection. In WT infection, the ratio of IE1 to IE2 was 1.5 at 2 days postinfection and then dropped to 0.4 ± 0.07 as IE2 levels increased and IE1 levels decreased over the remainder of the time course. In contrast, in UL135_STOP infection, the ratio of IE1 to IE2 was 32 at 2 dpi and dropped to an average of 3.6 ± 1.6 over the remainder of the time course. IE2 levels remained low and were not amplified at later times in infection with UL135_STOP, indicating a possible defect in progression from the early to late stages of infection. At higher MOIs (2.0, single-step growth), virus replication and protein accumulation were identical to those for the WT (Fig. 4E and F). Furthermore, at an MOI of 2, the ratio of IE1 to IE2 was identical over the time course between WT and UL135_STOP infection. From these results, the defect in replication observed following transfection of UL135_STOP BAC DNA or during infection at a low MOI was largely overcome by infection with equal numbers of PFU at high multiplicities, where 15-fold more genomes were present in the infecting inoculum.

The altered genome-to-PFU ratio of UL135_STOP indicates a defect in viral progeny (Fig. 4A). Defective particles may fail to enter or initiate infection properly, such that the defect can be overcome by increasing the number of genome-containing particles in the virus inoculum (Fig. 4C). This prompted us to analyze entry of viral genomes and subsequent viral genome amplification during infection. We infected fibroblasts at an MOI of 0.5 and quantified the viral genomes over a time course of infection beginning at 6 h postinfection (hpi) using real-time PCR with a probe to UL122 (IE2) (Fig. 5). The 6-hpi time point precedes viral DNA synthesis and, therefore, serves as a measure of the number of input genomes. Cells were treated with deoxyribonuclease I prior to DNA extraction so that only viral genomes delivered to the cell would be measured. Even though the virus inoculum contained 15-fold more genomes than the number in WT or UL138_STOP infection (Fig. 4A), a similar number of UL135_STOP genomes entered the cells relative to the number for WT infection (Fig. 5, inset). This result suggests that a large proportion of the UL135_STOP genome-containing particles does not enter the cell efficiently. These results are consistent with a defect in UL135_STOP particles.

FIG 5.

UL135 is largely dispensable for viral genome amplification in fibroblasts. Total DNA was isolated from MRC-5 cells infected (0.5 PFU per cell) with TB40/E WT, TB40/E-UL135_STOP, TB40/E-UL138_STOP, andTB40/E-5/8_STOP over a time course. The number of viral genomes relative to the level of GAPDH expression was quantified by qPCR using IE2- and GAPDH-specific primers. The viral genomes present in the cell at 6 hpi represent input genomes and are replotted in the inset for better visualization. Bars represent the averages of triplicate measurements from three independent experiments, and standard deviations are shown.

Our analysis of viral genome accumulation during the course of infection indicated a delayed accumulation of UL135_STOP genomes over the time course of infection in fibroblasts (Fig. 5). The WT infection resulted in a 43-fold increase in the number of genomes between 6 and 72 hpi and a 77-fold increase by 120 hpi. UL135_STOP exhibited a delay in viral genome synthesis, with UL135_STOP infection resulting in a 17-fold amplification of genomes of between 6 and 72 hpi. However, the number of UL135_STOP genomes increased 88-fold by 120 hpi. UL138_STOP exhibited no defect in genome amplification, increasing the number of input genomes by 46-fold by 72 hpi and 70-fold by 120 hpi. Infection with 5/8_STOP resulted in a level of genome amplification comparable to that achieved with infection with WT. Taken together, our results to this point suggest a defect in progression of the infection in fibroblasts when UL138 is expressed in the absence of UL135, which may contribute to problems in virus maturation. As many of the defects associated with UL135_STOP are only partially overcome in 5/8_STOP infection, UL135 may have independent functions that are not compensated for by the loss of UL138.

Virus maturation is defective in the absence of pUL135.

We were struck by the failure of UL135_STOP BAC DNA to initiate infection in transfected fibroblasts (Fig. 1), the altered genome-to-PFU ratio (Fig. 4), and the delay in viral genome synthesis (Fig. 5) when UL138 was expressed in the absence of UL135. To further investigate the possible defect in virus maturation and the interplay between UL135 and UL138, we evaluated progeny virus formed in fibroblasts infected with WT, UL135_STOP, UL138_STOP, and 5/8_STOP using transmission electron microscopy (TEM). Fibroblasts were infected with each virus at an MOI of 2 and processed for TEM at 2 dpi. Enveloped virions were present in all infections (Fig. 6). However, noninfectious, enveloped particles or NIEPs that lacked a viral genome were more prevalent in UL135_STOP-infected cells than WT virus-infected cells (Fig. 6A versus B). The production of progeny virus in cells infected with UL138_STOP (Fig. 6C) or 5/8_STOP (Fig. 6D) resembled that in cells infected with WT virus. Our analysis further revealed that in addition to increased frequencies of NIEPs, many cytoplasmic UL135_STOP genome-containing particles were malformed (Fig. 6E). Enveloped virions in UL135_STOP infection appeared to lack a full complement of tegument, and many virions appeared to be loosely enveloped.

FIG 6.

UL135_STOP infection exhibits defects in virion maturation. MRC-5 fibroblasts were infected with TB40/E WT (A), TB40/E-UL135_STOP (B and E), TB40/E-UL138_STOP (C), or UL135_STOP/UL138_STOP (D) at an MOI of 2. At 3 dpi, cells were fixed, embedded, and sectioned for transmission electron microscopy. Representative micrographs are shown. Black arrows, mature, genome-containing virions; black arrowheads, NIEPs; white arrows and white arrowheads, loosely enveloped and nonenveloped nucleocapsids in the cytoplasm, respectively. Magnifications, ×15,000; bars, 500 nm. (F) The ratio of infectious virions to NIEPs was scored in micrographs of 15 to 25 MRC-5 cells infected as indicated for panels A to D. Six hundred to 800 virions were counted for each infection. Standard deviations are shown. An unpaired, two-tailed Student's t test was used to determine significance for each mutant virus relative to the results for WT. *, P = 0.005. All other P values were >0.5.

We quantified the ratio of the number of normal genome-containing virions to the number of either NIEPs or abnormal genome-containing particles in electron micrographs of 15 to 25 cells for each virus infection (Fig. 6F). In WT, UL138_STOP, and 5/8_STOP infections, 4 mature virions were counted for every NIEP, whereas 2 mature virions were counted for every NIEP in UL135_STOP infection. Further, 8 to 10 normal virions were counted for every abnormal genome-containing virion in WT-, UL138_STOP-, and 5/8_STOP-infected cells. In contrast, less than 4 normal virions were counted for every abnormal genome-containing virion in UL135_STOP infection. The defects in UL135_STOP virion morphogenesis were largely resolved by the additional disruption of UL138 in 5/8_STOP infection. The rescue of these phenotypes in 5/8_STOP infection (Fig. 6D) is consistent with our previously reported TEM analysis, where no defect in the maturation of mature virions in fibroblasts infected with UL133/8_NULL virus was detected (42). While the standard deviations associated with this quantification are large and overlapping due to the variation in the number of virions present in each micrograph counted (range, 10 to 150), the differences observed for UL135_STOP infection relative to WT infection were statistically significant. The level of significance between WT and UL135_STOP achieved was due to the fact that abnormal virions outnumbered normal virions in many UL135_STOP micrographs. In the case of WT and UL138_STOP infection, abnormal virions (NIEPs or genome-containing virions) never outnumbered normal virions and only rarely outnumbered normal virions in 5/8_STOP infection.

We attempted to band the UL135_STOP virions present in cell-free supernatants from infected fibroblasts on glycerol-tartrate gradients. While WT, UL138_STOP, and 5/8_STOP virions were readily banded, UL135_STOP virions failed to band (data not shown). In the case of UL135_STOP, infectivity and viral DNA were distributed throughout the gradient. The inability to band UL135_STOP virus products was consistent with the heterogeneous nature of the virus particles observed in infected cells by TEM (Fig. 6).

UL135 is required for genome amplification and reactivation in CD34⁺ HPCs.

We have previously demonstrated a role for UL138 in suppressing virus replication to favor latency (23, 29). Viruses lacking UL138 replicate with greater efficiency than WT viruses in CD34⁺ HPCs. Given the opposing roles of UL135 and UL138 on virus replication, we hypothesized that pUL135 might be important for balancing states of latency and reactivation in CD34⁺ HPCs. The finding that UL135_STOP exhibited no defect for replication at a high MOI using equivalent numbers of PFU was fortuitous and allowed us to distinguish a defect in reactivation from a defect in replication. We infected CD34⁺ HPCs with WT, UL135_STOP, UL138_STOP, and 5/8_STOP at equivalent numbers of PFU (2 PFU/cell). Under these conditions, the virus replicates like the WT in fibroblasts (Fig. 4E and F). Infected CD34⁺ HPCs (GFP⁺) were isolated by fluorescence-activated cell sorting and seeded into long-term bone marrow culture medium over a stromal cell support. Total DNA was isolated at 2, 5, and 10 dpi, and viral genomes were quantified by qPCR. Despite the increased numbers of viral genomes in the UL135_STOP inoculum (Fig. 4A), the numbers of viral genomes present in infected cells at 2 dpi were within 2-fold of each other in each infection (Fig. 7A), indicating the delivery of a similar number of input genomes. Viral genomes were modestly amplified during infection in CD34⁺ HPCs. In the WT infection, the number of viral genomes increased by 5-fold at between 2 and 10 dpi. Although UL135_STOP genomes were maintained over the course of infection in CD34⁺ HPCs, the number of genomes increased by only 1.7-fold from 2 to 10 dpi in UL135_STOP infection. Consistent with the loss of the latency phenotype previously reported for UL138_STOP, virus genomes for this mutant were amplified 12-fold over the time course. These results indicate that UL135 promotes while UL138 restricts early stages of replication in CD34⁺ HPCs. Similar to the findings of WT infection in CD34⁺ HPCs, the number of viral genomes increased 5-fold in 5/8_STOP infection. The data derived from the 5/8_STOP double mutant virus indicate that UL135 is required for genome amplification only when UL138 is expressed.

FIG 7.

UL135 is required for reactivation from latency. (A) Total DNA was isolated from CD34⁺ HPCs over a time course following infection with TB40/E WT, TB40/E-UL135_STOP, TB40/E-UL138_STOP, and TB40/E-UL135_STOP/UL138_STOP at an MOI of 2. The number of viral genomes relative to the level of GAPDH expression was quantified by qPCR using IE2- and GAPDH-specific primers. The viral genomes present in the cell at 2 dpi represent input genomes. Bars represent the averages of triplicate measurements from three independent experiments. (B) Pure populations of CD34⁺ HPCs infected with TB40/E WT, TB40/E-UL135_STOP, TB40/E-UL138_STOP, or TB40/E-UL135_STOP/UL138_STOP at an MOI of 2 were isolated by fluorescence-activated cell sorting at 24 hpi and maintained in LTBMC medium. At 10 dpi, viable CD34⁺ HPCs were seeded onto MRC-5 cell monolayers plated in 96-well dishes by limiting dilution (reactivation). An equivalent number of cells was mechanically disrupted and seeded in parallel to determine the infectious virus present in the cultures prior to reactivation (prereactivation). The frequency of infectious center formation pre- and postreactivation was determined 14 days later from the number of GFP⁺ wells at each dilution using ELDA software. Bars represent the averages of three to six independent experiments, and standard errors are shown. Student's t test was used to determine the significance of the differences between prereactivation and reactivation for each virus infection. *, P = 0.02 for WT virus infection; P values were >0.1 for all mutant virus infections.

We next analyzed the infectious centers produced in CD34⁺ HPCs during the 10-day latency period in comparison to the infectious centers derived from reactivation (Fig. 7). For these experiments, viable infected cells were seeded by limiting dilution (96-well dishes) into a coculture with permissive fibroblasts in a cytokine-rich medium to favor reactivation of latent virus. To distinguish infectious centers formed from reactivation from those formed during the latency period prior to reactivation, we seeded a lysate prepared from an equivalent number of cells by limiting dilution in parallel. Fourteen days later, GFP⁺ wells were scored and the frequency of infectious centers was determined. As previously reported (27), WT established a latent infection, which could be reactivated to produce a 35-fold increase in the frequency of infectious centers (Fig. 7B) (P = 0.02). Similar to our previous reports, UL138_STOP exhibited a loss of the latency phenotype such that the virus replicated in CD34⁺ HPCs prior to the reactivation stimulus. Indeed, there was only a 1.6-fold increase following reactivation over the frequency of infectious centers produced prior to reactivation. Intriguingly, UL135_STOP reactivated poorly, producing a 5-fold greater frequency of infectious centers during reactivation than prior to reactivation. It should be noted that only 2 to 4 total infectious centers were counted following reactivation of UL135_STOP. The difference between the frequencies of infectious centers formed prior to and following reactivation was not significant (P = 0.17). The defect in reactivation in CD34⁺ HPCs is striking, since UL135_STOP replicates like WT virus in fibroblasts under these infection conditions (MOI, 2). Because viral genomes were maintained but not amplified during infection in UL135_STOP-infected cells, it is possible that the failure to reactivate is due to a defect in replicative events important during the establishment of latency. The 5/8_STOP infection resulted in an intermediate phenotype where reactivation was crippled in the absence of UL135 and the replication advantage prior to reactivation due to the lack of UL138 was no longer observed. Reactivation of cells infected with 5/8_STOP resulted in a 14-fold increase in the number of infectious centers relative to that for the lysate control, but this difference failed to reach statistical significance (P = 0.19). These results are consistent with a requirement for UL135 in viral replication and reactivation from latency.

DISCUSSION

Our work investigates the role of the ULb′-encoded UL135 gene during HCMV infection and reveals the existence of a novel molecular switch encoded by genes within the UL133/8 locus that balances latent and replicative states of infection in CD34⁺ HPCs. This is the first demonstration of a requirement for a ULb′ gene for replication in cultured fibroblasts and provides an avenue for new insights into the complex interplay between viral genes regulating states of virus infection. As disruption of the entire UL133/8 locus has modest effects on replication in fibroblasts, the extreme replication defect associated with the reconstitution of infection from viral genomes lacking UL135 was surprising. This finding suggested that antagonistic functions are encoded within the locus. Accordingly, the phenotypes associated with the disruption of UL135 were largely overcome by subsequent disruption of UL138. This finding indicates an intriguing relationship between UL135 and UL138 whereby UL138 suppresses virus replication for latency and UL135 overcomes this suppression to promote viral replication and reactivation (Fig. 8).

FIG 8.

Model of UL135 antagonism of the latency-promoting functions of UL138. Our work reveals a novel role for UL135 in overcoming the suppressive effects of UL138 to promote viral replication and reactivation in CD34⁺ HPCs. Our data suggest a role for UL135 in promoting the progression of infection beginning with an impact on IE2 expression and viral genome replication. We have detected the expression of pUL138, but not pUL135, in latently infected CD34⁺ HPCs. In fibroblasts where both UL135 and UL138 are expressed, UL135 is dominant and the suppressive effects of UL138 are subtle. In CD34⁺ HPCs, UL138 suppresses replication for latency and UL135 is required for reactivation from latency.

pUL135 was detected as three protein isoforms (Fig. 2). We demonstrate that the two primary isoforms originate from the first two methionine codons in the 308-amino-acid coding sequence (NCBI GenBank accession number ADE88119). The highest-molecular-mass isoform likely represents a posttranslational modification of the primary full-length isoform. This result indicates that, at least in fibroblasts, the methionine start codon in the 328-amino-acid annotation is not used. Previously, we determined that the UL133/8 locus is expressed as three large transcripts of 3.6, 2.7, and 1.4 kb (28). UL135 was predominantly expressed from the 2.7-kb transcripts, although UL135 could also be expressed from the 3.6-kb transcript. The transcriptional start site of the 2.7-kb transcript is just upstream of the initiating AUG of the 308-amino-acid annotation and does not include the initiating AUG of the 328-amino-acid annotation. Consistent with our rapid amplification of cDNA ends experiments (28), ribosomal profiling experiments detect mRNAs encoding the 308-amino-acid pUL135 (43). Taken together, these studies and our data presented herein suggest that UL135 encodes a 308-amino-acid protein. However, these studies do not rule out the possibility of expression of the 328-amino-acid protein from the 3.6-kb transcript in other cell types or under other infection conditions. Further, for the first time, we detected the existence of a smaller isoform of pUL135 (Fig. 2A and 3B), which is translated from a downstream initiation codon that would result in a 223-amino-acid protein. Unlike the other two isoforms, this smaller isoform is not predicted to be membrane associated. The relative levels of the three pUL135 isoforms vary with respect to time during infection (Fig. 3B). The presence of multiple protein isoforms derived from multiple transcripts is consistent with the complex nature of translation from proteins within this region (28; K. Caviness and F. D. Goodrum, unpublished results). Further work is required to characterize each of these isoforms and their relative contributions to infection.

Disruption of UL135 results in multiple, complex phenotypes. Viruses lacking only UL135 in either the FIX or TB40/E strain were difficult to reconstitute following transfection of BAC clones of the viral genome into permissive fibroblasts (Fig. 1). At lower MOIs, UL135-deficient viruses were delayed in immediate early gene expression and produced altered ratios of the major immediate early proteins IE1 (72 kDa) and IE2 (86 kDa) (Fig. 4D). These observations indicate defects in the progression of very early stages in the lytic cycle. Defects in immediate early gene expression are unexpected, as the transcripts encoding pUL135 are expressed with early kinetics (29) and pUL135 accumulates in the early phase (27). As neither pUL135 nor the other UL133/8 proteins are incorporated into the virion (Fig. 2C), these proteins cannot promote IE gene expression or virus replication as a constituent of the tegument.

The altered ratio of IE1 to IE2 may indicate a dysregulation of transcription, RNA splicing, or protein turnover during the immediate early and early phases of infection. IE2 is a potent and promiscuous transactivator expressed in the earliest stages of infection to establish viral replication (44), and its levels are tightly regulated by a negative-feedback mechanism (45–48). At the onset of viral DNA synthesis, IE2 (but not IE1) expression is amplified during a second wave of early gene expression (49–51). Our data show that in the absence of UL135, amplification of IE2 expression coincident with viral DNA synthesis does not occur. Consistent with the diminished expression of IE2 in UL135_STOP infection, early and late gene expression is delayed and, in some cases, reduced (Fig. 4D). Further, viral genome amplification (Fig. 5) is also delayed. It is not clear if these later defects are due to the absence of UL135 or simply the cumulative cost of altered IE gene expression (45). Interestingly, defects in gene expression and replication associated with UL135_STOP infection were overcome by increased multiplicities of infection (Fig. 4E and F). The impact of UL135 on the IE and early stages of infection is the focus of our ongoing investigation.

The defects in virus maturation associated with disruption of UL135 were striking (Fig. 4 and 6). While these defects may be secondary to defects in stages of infection preceding maturation, the defects are multilayered. Fibroblasts infected with UL135-deficient virus produced a greater proportion of NIEPs and defective genome-containing virus particles with abnormal envelopes that appeared to lack tegument (Fig. 6). We recently demonstrated that while the UL133/8 locus was required for virus maturation in endothelial cells, a virus lacking the entire UL133/8 locus, UL133/8_NULL, exhibited no defects for virus maturation in fibroblasts (42). However, the results reported here suggest that UL135 is required for maturation when UL138 is expressed. Further work is required to understand cell type-dependent functions and how UL135 and UL138 impact the formation of genome-containing virions and cytoplasmic maturation. Studies into the interplay between UL135 and UL138 may contribute unique insights into the relationship between defects in the immediate early and early stages of infection (e.g., IE2 expression and viral genome synthesis) and DNA packaging and virus maturation.

In CD34⁺ HPCs, the defects in viral genome synthesis and replication associated with UL135_STOP infection were more pronounced, even at multiplicities of infection that resulted in WT replication in fibroblasts. While viral genomes were maintained in UL135_STOP-infected cells, they were not amplified and were unable to support virus replication/reactivation (Fig. 7). From these data, we propose that the UL135_STOP infection represents a nonreactivatable, latent state. Our data for both fibroblasts and CD34⁺ HPCs suggest a role for pUL135 in promoting the progression of very early stages of infection. Therefore, it is possible that pUL135 is important for events early during infection in CD34⁺ HPCs to establish latency, in the reactivation of productive replication from latency, or both.

The defects in both gene expression and maturation associated with the loss of UL135 could be largely overcome by the disruption of UL138, suggesting that UL138 imposes a restriction to viral replication at very early points in replication that UL135 is required to counteract for reactivation (Fig. 8). Consistent with this notion, viruses lacking UL138 synthesize increased levels of IE2, early, and late transcripts in fibroblasts (52) and increased levels of viral DNA in both fibroblasts (52) and CD34⁺ cells (Fig. 7A). Because some defects associated with UL135_STOP are only partially overcome by disruption of UL138, UL135 may also have independent functions that are not compensated for by the loss of UL138. It will be important to define both the UL138-related and -independent functions of pUL135 in understanding the interplay between pUL135 and pUL138. We previously reported that UL135 and UL138 are both abundantly expressed during productive viral replication in fibroblasts and endothelial cells. However, pUL138, but not pUL135, is detected during latency in CD34⁺ HPCs (27). The discordant expression of UL138 and UL135 in CD34⁺ HPCs might confer an advantage to replication-suppressive pUL138 favoring the establishment of latency. It is not known how pUL135 and pUL138 expression may be differentially regulated from the locus in a cell type-dependent manner. The UL133/8 locus is encoded on 3′ coterminal transcripts of 3.6, 2.7, and 1.4 kb (28). pUL135 can conceivably be expressed from the 3.6- and 2.7-kb transcripts, while pUL138 is expressed from all three transcripts and is expressed most abundantly from the 1.4-kb transcript (28). Differences in the relative abundance of these transcripts in CD34⁺ HPCs could result in differential expression of the proteins. However, the 3.6-, 2.7-, and 1.4-kb transcripts are present in CD34⁺ HPCs at a ratio of 1:3:3 (27), suggesting that the failure to detect pUL135 in CD34⁺ HPCs does not reflect lower levels of transcripts encoding UL135 relative to the levels of transcripts encoding UL138. It remains possible that the synthesis and turnover of pUL135 are differentially modulated in a cell type-dependent fashion. The molecular mechanisms by which UL135 and UL138 expression is regulated and how UL135 and UL138 antagonize one another are a primary focus in our laboratory.

Comparing the latency phenotypes of UL135_STOP and UL138_STOP to the latency phenotype of 5/8_STOP illustrates several important points. While the 5/8_STOP mutant virus exhibits replication (Fig. 4 and 5) and latency (Fig. 6 and 7) phenotypes that are restored from the phenotype of UL135_STOP, the levels of replication or reactivation from latency of this virus are not fully complemented to WT levels. The 5/8_STOP virus has lost the replication advantage of viruses lacking only UL138 and is not as restricted for reactivation as the virus lacking only UL135. The double mutant phenotype indicates that the phenotypes exhibited by UL135_STOP and UL138_STOP viruses are not only the result of the gene disrupted but also the result of the gene or genes that remain. For example, the defects in reactivation associated with UL135_STOP in CD34⁺ HPCs are largely due to the loss of the replication-promoting activities of UL135 in the presence of the replication-suppressive activities of UL138, rather than simply the loss of UL135. Similarly, the enhanced replication and enhanced viral DNA synthesis of UL138_STOP in CD34⁺ HPCs (loss of the latency phenotype) are due not only to the loss of the suppressive effects of UL138 but also to the presence of the replication-promoting activities of UL135. In the case of 5/8_STOP, lacking the activator and the suppressor, the virus retains a basal ability to respond to a reactivation stimulus. It is of interest to determine how UL133 and UL136, which remain intact in each of these mutant viruses, might contribute to these phenotypes.

Infection phenotypes associated with the chemical inhibition or genetic disruption of the UL97 viral kinase are strikingly similar to those resulting from the disruption of UL135. UL97 mutants exhibit a severe growth defect (53), which is greater in clinical strains than in laboratory-adapted strains of HCMV (54). As demonstrated by our collaborative work in the accompanying article (52), UL97 promotes the second phase of IE2 expression, viral DNA synthesis, and progression of the lytic infection cycle. This phenotype exhibits a striking resemblance to that of UL135_STOP (Fig. 4D). Interestingly, the function of UL97 in promoting replication during infection with clinical strains largely depends on the presence of UL135. Further, UL97 mutant viruses exhibit defects in the late stages of virus maturation, producing a number of empty virions and defects in secondary envelopment (55–58). The similarities in phenotypes associated with the loss of UL135 or UL97 and the interplay between UL135 and UL97 during infection that these phenotypes suggest are intriguing.

The dual requirement for pUL135 in replication/reactivation and virus maturation is reminiscent of functions associated with ICP0 or VP16 of herpes simplex virus 1 (HSV-1). Mutations in ICP0, the major transactivator of HSV-1, resulted in a 10-fold increase in the genome-to-PFU ratio relative to that for WT HSV-1 (59). Similarly, mutations in VP16 that disrupt its transactivating function result in a 100-fold increase in the genome-to-PFU ratio relative to the ratio for the WT (60) and corresponding defects in viral envelopment and maturation (61). ICP0 has a well-established role in HSV-1 reactivation from latency (62, 63). VP16 was recently shown to be required for acute replication in trigeminal ganglia and reactivation from latency (64). Therefore, defects in virus maturation and egress are common phenotypes associated with the disruption of viral genes important to reactivation. Although the prominent isoform of pUL135 is membrane associated (27), UL135 mutant viruses share striking similarities with ICP0 and VP16 mutant viruses, including (i) impaired replication at low multiplicities of infection by BAC transfection (60, 65), an (ii) an increased genome-to-PFU ratio (60, 65), and (iii) the cell cycle dependence of the severity of the defect. With regard to the last point, there is a reduced requirement for ICP0 for HSV-1 replication in cells arrested at G₁/S phase (66, 67). Since UL135_STOP virus replication from BAC transfection could be complemented under conditions with low levels of serum, UL135 may also resemble ICP0 is this regard. It is not yet known whether HCMV infection in the absence of UL135 but the presence of UL138 depends on the cell cycle or some other effect on the cellular environment induced by low levels of serum. Additional studies are required to understand how UL135 functions in the early stages of viral replication/reactivation, how the requirement for UL135 can be overcome in the absence of serum, and how its function contributes to productive replication and virus maturation.

Our study demonstrates the existence of opposing HCMV functions within the UL133/8 genetic locus, which we have previously shown modulate the outcome of viral infection in a cell type-dependent manner (27). From our studies, we propose a model (Fig. 8) whereby pUL135 and pUL138 comprise a molecular switch to balance states of latency and replication. In this model, pUL135 expression antagonizes pUL138-mediated suppression of early events in virus replication and appears to dominantly promote viral reactivation/replication. We have not detected a direct interaction between pUL135 and pUL138 (30), suggesting that the mechanism by which pUL135 relieves pUL138-mediated suppression is indirect. It is intriguing to speculate how these proteins may indirectly target one another through their interaction with cellular factors. While pUL138 has been shown to increase the levels of surface TNFR (31, 32) and decrease the levels of surface MRP-1 (33), pUL135 has not been implicated in the counterregulation of these proteins. Our data suggest an elegant mechanism by which the virus coordinately regulates the expression of two determinants from a single genetic locus which oppose one another with respect to their effects on viral replication. To our knowledge, this work defines the first mechanism balancing positively and negatively acting viral factors to control states of HCMV replication and latency.