JOURNAL OF VIROLOGY, Nov. 2010, p. 11661–11669 Vol. 84, No. 220022-538X/10/$12.00 doi:10.1128/JVI.00878-10Copyright © 2010, American Society for Microbiology. All Rights Reserved.

The Varicella-Zoster Virus ORFS/L (ORF0) Gene Is Requiredfor Efficient Viral Replication and Contains an Element

Involved in DNA Cleavage�

Benedikt B. Kaufer,1 Benjamin Smejkal,1 and Nikolaus Osterrieder1,2*Department of Microbiology and Immunology, College of Veterinary Medicine, Cornell University, Ithaca, New York 14853,1

and Institut fur Virologie, Freie Universitat Berlin, Philippstrasse 13, 10115 Berlin, Germany2

Received 23 April 2010/Accepted 3 September 2010

The genome of varicella-zoster virus (VZV), a human alphaherpesvirus, consists of two unique regions,unique long (UL) and unique short (US), each of which is flanked by inverted repeats. During replication, fourisomers of the viral DNA are generated which are distinguished by the relative orientations of UL and US. VZVvirions predominantly package two isomeric forms of the genome that have a fixed orientation of UL. An openreading frame (ORF) of unknown function, ORFS/L, also referred to as ORF0, is located at the extremeterminus of UL, directly adjacent to the a-like sequences, which are known to be involved in cleavage andpackaging of viral DNA. We demonstrate here that the ORFS/L protein localizes to the Golgi network ininfected and transfected cells. Furthermore, we were able to demonstrate that deletion of the predicted ORFS/Lgene is lethal, while retention of the N-terminal 28 amino acid residues resulted in viable yet replication-impaired virus. The growth defect was only partially attributable to the expression of the ORFS/L product,suggesting that the 5� region of ORFS/L contains a sequence element crucial for cleavage/packaging of viralDNA. Consequently, mutations introduced into the extreme 5� terminus of ORFS/L resulted in a defect in DNAcleavage, indicating that the region is indeed involved in the processing of viral DNA. Since the sequenceelement has no counterpart at the other end of UL, we concluded that our results can provide an explanationfor the almost exclusive orientation of the UL seen in packaged VZV DNA.

Varicella-zoster virus ([VZV] Human Herpesvirus 3), is ahighly cell-associated alphaherpesvirus that causes chickenpox (varicella) upon infection of naïve individuals (2). Dur-ing primary infection, VZV is able to establish latency incranial nerves, as well as dorsal root and autonomic ganglia,where it remains dormant until a reactivation event occurs(11). Reactivation of VZV occurs primarily in elderly or im-munocompromised individuals and results in the developmentof shingles (herpes zoster), which is often associated with se-vere pain and postherpetic neuralgia (1).

The VZV genome, the smallest among the human herpes-viruses, is approximately 125 kbp in size and encodes at least 70unique open reading frames (ORFs) (1). As has been reportedfor all alphaherpesviruses, the VZV genome consists of twounique regions, unique long (UL) and unique short (US), eachflanked by inverted repeat regions (TRL, IRL, TRS, and IRS)(9). In contrast to herpes simplex virus type 1 (HSV-1), theprototype alphaherpesvirus, VZV contains only very short re-peats (88 bp) on either end of UL, characteristic of members ofthe Varicellovirus genus (6). During alphaherpesvirus replica-tion, four isomers of viral DNA are generated which can bedistinguished by the orientation of UL and US relative to eachother. While all four possible isomers of HSV-1 DNA arepackaged in virions as equimolar populations, virions producedby VZV and other varicelloviruses, such as equine herpesvirustype 1 (EHV-1), contain predominantly only two of the four

possible isomeric forms of the genome (6, 10, 12, 15, 23). It wasshown by Southern blot analysis of VZV virion DNA thatinversion of the UL region is rare and occurs in only approxi-mately 5% of cases (6), which also may be attributed to a rarecircular configuration of the genome within the virion (14). Aprevious report on EHV-1 suggested that inversion of the UL

region in infected cells is common but that packaging occurs ina directional manner (23). For both VZV and EHV-1, thereason for the more-or-less exclusive orientation of UL withinthe virion still remains unknown.

The organization of the VZV genome is similar to that ofHSV-1, and over 90% of the VZV ORFs have counterparts inthe HSV-1 genome (1, 13). One of the genes with a predictedHSV-1 homologue is ORFS/L, also referred to as ORF0.ORFS/L is predicted to encode a tail-anchored 157-amino-acid(aa) residue type 2 transmembrane protein and was discoveredby Kemble and coworkers (13). The gene is located at the verybeginning of UL, directly adjacent to the a-like sequences thatcontain PacI and PacII sites crucial for the cleavage and packag-ing of concatameric VZV DNA (Fig. 1) (13, 20). Although nofunction has yet been attributed to the ORFS/L (ORF0) gene orits product, bioinformatic analysis of the VZV genome indicatedthat it represents a homologue of HSV-1 UL56 (RefSeq acces-sion no. NC_001348) (7, 8). While UL56 is dispensable for HSV-1replication in vitro, it plays an important role in pathogenicity invivo (3, 21). Little is known about the molecular mechanism ofUL56 function in the case of HSV-1, but UL56 orthologues arespecified by most members of the Alphaherpesvirinae subfamily(26). It was shown that the HSV-2 UL56 product localizes to theGolgi network and interacts with KIF1A, a kinesin motor protein,suggesting a role in vesicular trafficking (16, 17).

* Corresponding author. Mailing address: Institut fur Virologie,Freie Universitat Berlin, Philippstrasse 13, 10115 Berlin, Germany.Phone and fax: 49-30-2093-6564. E-mail: no.34@fu-berlin.de.

� Published ahead of print on 15 September 2010.

11661

A previous study of Kemble and coworkers also addressedthe localization of the ORFS/L protein using a rabbit poly-clonal antibody. It was reported that the ORFS/L product wasfound exclusively in the cytoplasm, which is contradictory tothe findings for the HSV-2 orthologue and also to the local-ization of the ORFS/L protein based on in silico predictionsfrom the primary sequence (13). ORFS/L of the P-Oka strainwas recently shown to be unglycosylated but present in thevirion (18). Furthermore, ORFS/L expression was detected inskin lesions of individuals, as well as neurons of dorsal rootganglia, during virus reactivation (13). In addition, the deletionof aa 29 to aa 157 of ORFS/L was shown to have an effect onviral replication in vitro and in vivo in the SCID-hu mousemodel with thymus-liver implants. In this study, a virus-en-coded luciferase reporter system was used to evaluate thegrowth properties of several bacterial artificial chromosome(BAC)-derived VZV mutants (28). However, it has remainedunknown whether the observed growth defect is dependent onORFS/L gene function or is due to the deletion of anothercritical sequence element.

In this study, we sought to perform a systematic analysisof ORFS/L sequences. We were able to demonstrate that theORFS/L protein localizes to the Golgi network in infected andtransfected cells, providing further evidence for its predicted

structure as a tail-anchored type 2 transmembrane protein andlending further support to the notion that it is the orthologueof HSV UL56. In addition, we showed that the ORFS/L geneproduct is important for efficient VZV replication in vitro. How-ever, we also identified a 5� region of the predicted ORFS/L thatis essential for replication and plays a role in cleavage of viralDNA, as previously suggested by Davison and colleagues (6,7). Since this essential region is not present at the opposite endof UL, it could provide an explanation for the almost exclusivepackaging in VZV virions of two viral DNA isomers with aninvariable UL orientation.

MATERIALS AND METHODS

Cells and viruses. Human melanoma cells (MeWo) were grown and main-tained in growth medium (minimal essential medium [MEM] supplemented with10% fetal bovine serum [FBS], 100 U/ml penicillin, and 0.1 mg/ml streptomycin)at 37°C under a 5% CO2 atmosphere (5). Wild-type and mutant viruses werereconstituted, grown, and amplified on MeWo cells by cocultivation of infectedwith uninfected cells at a ratio of 1:2 to 1:5. Virus stocks were frozen in growthmedium supplemented with 8% dimethyl sulfoxide (DMSO) and stored in liquidnitrogen.

Generation of ORFS/L mutants and revertants via en passant mutagenesis.Mutants with a deletion of ORFS/L (�aa1-157), a start codon knockout(aa1stop) in which the ATG was replaced by a stop codon (TAG), a C-terminaldeletion (�aa34-157), a C-terminal hemagglutinin (HA)-tagged ORFS/L(S/L_cHA), or a stop codon insertion at amino acid position 3 (aa3stop, TTT to

FIG. 1. Overview of the VZV ORFS/L genomic region and the mutants generated. (A) Schematic representation of the VZV genome with afocus on the terminal region containing ORFS/L. Scale bars provide an accurate measure of the genome and the expanded region. (B) Overviewof the mutants generated with mutations in the ORFS/L region. A cross indicates the deletion of the corresponding region. Black arrows indicatethe loci of stop codon or HA tag insertion.

11662 KAUFER ET AL. J. VIROL.

TAA) or 34 (aa34stop, TAC to TAG) were generated based on pP-Oka, aninfectious BAC clone of the P-Oka strain (24), via two-step Red-mediated enpassant mutagenesis as described previously (25). The HA tag was also insertedat the C terminus of the aa1stop, aa3stop, and aa34stop mutant viruses asdescribed above, resulting in aa1stop-cHA, aa3stop-cHA, and aa34stop-HA. Inorder to generate revertants of the respective ORFS/L mutants, a transfer vectorwas generated. Briefly, a region containing ORFS/L and flanking sequences wasamplified from pP-Oka by PCR, cloned into the pcDNA3.1 TOPO vector (In-vitrogen), and termed prORFS/L. The aphAI-I-SceI cassette was amplified frompEPkan-SII (25) and inserted into prORFS/L using a unique EcoRI restrictionsite within ORFS/L, resulting in plasmid prORFS/L-aphAI-I-SceI. The entirerORFS/L-aphAI-I-SceI cassette was amplified by PCR and used for two-stepRed-mediated mutagenesis. Revertant viruses of �aa1-157, aa1stop, and �aa34-157 were termed �aa1-157-rev, aa1stop-rev, and �aa34-157-rev, respectively. Allclones were confirmed by DNA sequencing of the ORFS/L region (Fig. 1A), aswell as multiple restriction fragment length polymorphism analyses (RFLP), toensure the integrity of the genome (data not shown). All oligonucleotides usedare given in Table 1.

Reconstitution of VZV from BAC DNA. BAC DNA used for transfection wasisolated with the Qiagen Midiprep system according to the manufacturer’s(Qiagen) instructions. Subsequently, MeWo cells were transfected with Li-pofectamine 2000 (Invitrogen) as described previously (24). Briefly, 1 �gBAC DNA was cotransfected with 200 ng pCMV62, a plasmid which containsthe VZV immediate-early (IE) gene ORF62 under the control of the major IE

promoter of human cytomegalovirus (CMV) (kindly provided by Kip Kinching-ton, University of Pittsburgh Medical School). The DNA-Lipofectamine solutionwas added to the cells and incubated for 4 h. The transfection medium was thenreplaced with growth medium. The resulting viruses were termed v�aa1-157,v�aa1-157-rev, vaa1stop, vaa1stop-rev, v�aa34-157, v�aa34-157-rev, vaa3stop,vaa34stop, vS/L_cHA, vaa1stop-cHA, vaa3stop-cHA, and vaa34stop-cHA.

Multistep growth kinetics and plaque size assay. One million MeWo cellswere inoculated with 100 PFU of cell-associated viruses. For multistep growthkinetics, infected cells were trypsinized at 24, 48, 72, 96, or 120 h postinfection(p.i.) and titrated in 10-fold dilutions with fresh MeWo cells, and the number ofplaques was determined by indirect immunofluorescence (IIF) using an anti-gBantibody (United States Biological). Three independent experiments were per-formed in triplicate. For plaque size assays, cells were fixed 6 days p.i. and stainedand plaques imaged with a digital camera (Zeiss Axiovert 25 and Axiocam).Areas of 60 individual plaques were evaluated with Axiovision software (Zeiss).

Generation of ORFS/L expression plasmids. ORFS/L was amplified frompP-Oka and inserted into pcDNA3.1/V5-His TOPO vector (Invitrogen). Theresulting plasmid was sequenced and termed pORFS/L-His. It was then trans-fected with Lipofectamine 2000 according to the manufacturer’s instructions(Invitrogen).

Indirect immunofluorescence labeling and confocal microscopy. MeWo cellswere fixed with 3% paraformaldehyde (PFA) and permeabilized with 0.1%saponin at 48 h after infection or transfection. Cells were stained with mousemonoclonal anti-GM130 antibody (BD Transduction Laboratories), rabbit poly-

TABLE 1. Primers used for cloning and mutagenesis

Construct Direction Sequence (5� 3 3�)a

�aa1-157 Forward CCCACCTCCCCGCGCGTTTGCGGGGCGACCATCGGGGGGGTGAAACTACTGTCCGGAAGGGTAGGGATAACAGGGTAATCGATTTATTC

Reverseerse GCAAGCGAGAATAAATACCTTCCCCTTCCGGACAGTAGTTTCACCCCCCCGATGGTCGCCCCGCGCCAGTGTTACAACCAATTAACC

�aa1stop Forward CCCCACCTCCCCGCGCGTTTGCGGGGCGACCATCGGGGGGGTAGGGGATTTTTTGCCGGGAAACCTAGGGATAACAGGGTAATCGATTTA

Reverse GTTAAAGGCTGGCGGGGGGGGTTTCCCGGCAAAAAATCCCCTACCCCCCCGATGGTCGCCCCGCGCCAGTGTTACAACCAATTAACCAA

�aa34-157 Forward GCGTCCACCCCTCGTTTACTGCTCGGATGGCGACCGTGCACTGAAACTACTGTCCGGAAGGGGTAGGGATAACAGGGTAATCGATTT

Reverse GCAAGCGAGAATAAATACCTTCCCCTTCCGGACAGTAGTTTCAGTGCACGGTCGCCATCCGAGCGCCAGTGTTACAACCAATTAACC

�aa3stop Forward CCCGCGCGTTTGCGGGGCGACCATCGGGGGGGATGGGATAATTTGCCGGGAAACCCCCTAGGGATAACAGGGTAATCGATTT

Reverse GGGTTTTGTTAAAGGCTGGCGGGGGGGGTTTCCCGGCAAATTATCCCATCCCCCCCGATGGTCGCCGCCAGTGTTACAACCAATTAACC

�aa34stop Forward GTCCACCCCTCGTTTACTGCTCGGATGGCGACCGTGCACTAGTCCCGCCGACCTGGGACCCTAGGGATAACAGGGTAATCGATTT

Reverse GACGACGTGAGGGTGACCGGCGGGGTCCCAGGTCGGCGGGACTAGTGCACGGTCGCCATCCGGCCAGTGTTACAACCAATTAACC

S/L_cHA Forward CCGTTTTTCCCGAGGAACCTCCCAACTCAACTACATATCCGTATGATGTGCCGGATTATGCGTGAAACTAGGGATAACAGGGTAATCGAT

Reverse GAATAAATACCTTCCCCTTCCGGACAGTAGTTTCACGCATAATCCGGCACATCATACGGATATGTAGGCCAGTGTTACAACCAATTAACC

prORFS/L Forward AGCGACCCCACCTCCCCReverse CGACAAGCTGCAAGCGAGAA

prORFS/L-aphAI-I-SceI Forward CTAAGAATTCGCAAATGCTGTGTACCGGCTAGGGATAACAGGGTAATCGATTTReverse TTGCGAATTCTTAGAAAAGCCAGCTGAAGTCTGGCCAGTGTTACAACCAATTAACC

pCDNA-ORFS/L Forward ATGGGATTTTTTGCCGGGReverse TGTAGTTGAGTTGGGAGGTTCCTC

Southern probe VZV Forward GACGCACCGGGGTCATCterminus Reverse TGTAGTTGAGTTGGGAGGTTCCTC

a Underlined sequences indicate restriction enzyme sites. Bold indicates mutated sequences.

VOL. 84, 2010 VZV ORFS/L (ORF0) AND REPLICATION 11663

clonal anti-His antibodies (Genscript Cooperation), or rabbit polyclonal anti-HAantibodies (Zymed Laboratory) at 1:200 dilutions. Secondary antibodies, includ-ing Alexa Fluor 568-conjugated goat anti-mouse and Alexa Fluor 488-conjugatedgoat anti-rabbit (Molecular probes), were diluted 1:1,000. Cells were washedwith phosphate-buffered saline (PBS) and mounted with 4�,6�-diamidino-2-phe-nylindole (DAPI) Vectashield (Vector Laboratories). Cells were examined usingan SP5 confocal microscope system (Leica). Collected images were arrangedusing Photoshop 7.0 (Adobe).

Western blot assay. MeWo cells were infected with vP-Oka, vS/L_cHA,vaa1stop-cHA, vaa3stop-cHA, or vaa34stop-cHA. Infected cells were harvestedafter infection for 72 to 96 h, separated by sodium dodecyl sulfate (SDS)–15%polyacrylamide gel electrophoresis (PAGE), and analyzed by Western blottingusing a polyclonal rabbit anti-HA (Zymed Laboratory) or a monoclonal rabbitanti-�-actin (Cell Signaling Technology) antibody, each at a 1:1,000 dilution.Target proteins were visualized using an anti-rabbit peroxidase-conjugatedantibody (Sigma-Aldrich) and enhanced chemiluminescence (ECL; Pharmacia-Amersham), and the signal was recorded on X-ray film (Amersham Biosciences)as described previously (22).

Cleavage assay. Ten million MeWo cells were infected with 1 � 104 PFU ofvP-Oka, vaa1stop, vaa1stop-rev, vaa3stop, or vaa34stop, harvested when com-plete cytopathic effect was observed (72 h for the wild type and 96 h forvaa1stop), and documented via bright-field microscopy. Cells were washed withice-cold PBS and lysed in buffer A (10 mM Tris, pH 7.4, 10 mM NaCl, 3 mMMgCl2, 0.5% NP-40). Either purified nuclei or whole-cell lysates were digestedwith proteinase K (Qiagen) in the presence of 1% SDS and 5 mM EDTA, andDNA was purified by phenol-chloroform extraction (22). Equivalent amounts ofpurified nuclear or whole-cell DNA were digested with BamHI overnight andseparated by agarose gel electrophoresis. Southern blot analysis allowed thedetection of VZV terminal fragments using a digoxigenin (DIG)-labeledprobe, generated with a PCR DIG synthesis kit (Roche) according to themanufacturer’s instructions. The oligonucleotides used for amplification aregiven in Table 1. Southern blot signals were detected by chemiluminescenceas recorded with a ChemiGenius bioimaging system and quantified withGeneTool software (Syngene, Cambridge, United Kingdom).

RESULTS

Analysis of ORFS/L in infected cells. ORFS/L, predicted toencode a protein of 157 aa in length, harbors two alternativestart codons that could initiate translation at aa 29 or aa 51,respectively. In order to identify protein species generated bythe ORFS/L open reading frame, we inserted an HA tag at theC terminus (pS/L_cHA) in pP-Oka, an infectious BAC clone ofthe P-Oka strain (Fig. 1A and B) (24). Multistep growth ki-netics and plaque size assays demonstrated that S/L_cHA virus(vS/L_cHA) is capable of replicating with kinetics comparableto those of parental BAC-derived vP-Oka, indicating that HAtag insertion had no effect on viral replication in vitro (Fig. 2Aand B). Western blot analysis of vS/L_cHA-infected MeWocells revealed that ORFS/L is expressed as a single species ofabout 18 kDa, corresponding well to the predicted molecularmass of the full-length, 157-aa protein (Fig. 2C).

ORFS/L localizes to the trans-Golgi network in infected andtransfected cells. To determine the subcellular localization ofORFS/L in VZV-infected cells, we infected MeWo cells withvS/L_cHA. Confocal microscopy revealed that ORFS/L colo-calizes with GM130, a marker for the Golgi apparatus, ininfected MeWo cells (Fig. 3A, upper panel). ORFS/L alsolocalized to the Golgi network in virus-induced syncytia, as wellas adjacent infected cells (Fig. 3A, middle panel). In additionto that, ORFS/L was also localized to the nuclear envelopeupon syncytium formation (Fig. 3A, middle and lower panels).To address the question of whether the localization of ORFS/Lto the Golgi network is dependent on other viral proteins, wedetermined the localization of ORFS/L in transfected cells.MeWo cells were transfected with pcDNA-ORFS/L-His. Sim-

ilar to the situation in infected cells, ORFS/L was localized tothe Golgi network after expression of the His-tagged proteinunder a strong herpesvirus (CMV-IE) promoter. We con-cluded, therefore, that the localization of the ORFS/L proteinto the Golgi apparatus is not dependent on the presence ofother viral proteins (Fig. 3B) and that its presence at cellularmembranes is consistent with in silico predictions suggestingthat ORFS/L represents an orthologue of HSV-1 UL56.

Deletion of the entire ORFS/L is lethal for VZV replicationin vitro. Previously, a recombinant virus containing a deletionof the C-terminal 129 aa of the 157-aa protein downstream ofthe second start codon of ORFS/L was reported to have aneffect on viral replication in vitro and in vivo (28). To confirmthe role of ORFS/L in VZV replication, we deleted the entirepredicted ORF (aa 1 to aa 157), corresponding to nucleotidepositions 88 to 558 of pP-Oka (Fig. 1) (24). In addition, a

FIG. 2. Growth properties of vS/L_cHA and size determination ofORFS/L. (A) Multistep growth kinetics of vS/L_cHA and vP-Oka.Shown are the means and standard errors (error bars) of the results ofthree independent experiments. (B) Plaque size measurements of vS/L_cHA and vP-Oka. Shown are relative average plaque areas of 60individual plaques with standard deviations (error bars), with the sizesof vP-Oka plaques set at 100%. (C) Western blot analysis of cellsinfected with vP-Oka (lane 1) or vS/L_cHA (lane 2). The arrow indi-cates the full-length ORFS/L protein, which has the predicted molec-ular mass.

11664 KAUFER ET AL. J. VIROL.

revertant BAC clone (p�aa1-157-rev) was generated in whichthe original ORFS/L sequence was restored. Subsequently, theconstructs were transfected into MeWo cells, resulting in thereconstitution of recombinant viruses v�aa1-157 and v�aa1-157-rev. Upon virus reconstitution, two independent v�aa1-157 clones exhibited a severe replication defect, while the rever-tant virus grew with kinetics that were virtually indistinguishablefrom those of the parental vP-Oka. Plaque size measurementsconfirmed this significant growth defect (Fig. 4A and B). Thereplication defect of both v�aa1-157 clones was so severe thatthe viruses could not be amplified by passaging of the virus.

ORFS/L gene product plays a role in replication in vitro. Todetermine whether the observed growth effect was due to thelack of the ORFS/L gene product or an essential DNA elementwithin ORFS/L, we generated a series of additional mutantviruses. In order to abrogate ORFS/L gene expression, wereplaced the 1st start codon (vaa1stop), the codon for aminoacid 3 (vaa3stop), or the codon for amino acid 34 (vaa34stop),just downstream of the 2nd start codon, with a stop codon (Fig. 1).We also generated a C-terminal deletion that results in the ab-sence of aa 34 to aa 157 (v�aa34-157). The latter mutant is verysimilar to two mutants generated in previous reports and served

as a control (13, 28). In addition, revertant viruses of vaa1stop(vaa1stop-rev) and v�aa34-157 (v�aa34-157-rev) were generated.Multistep growth kinetics revealed that vaa1stop, vaa3stop,vaa34stop, and v�aa34-157 were significantly impaired in viralreplication compared to vP-Oka or revertant viruses (Fig. 5A).While vaa3stop, vaa34stop, and v�aa34-157 replication wasreduced by about 5-fold, the replication of vaa1stop was re-duced more and resulted in an approximately 10-fold reduc-tion of virus progeny at all time points analyzed. In order toconfirm the differences in growth defects, we also performedplaque size assays. In agreement with the multistep growth kinet-ics, vaa1stop, vaa3stop, vaa34stop, and v�aa34-157 showed sig-nificant defects in plaque formation compared to that of vP-Okaor revertant viruses (Fig. 5B). The mean plaque areas of vaa1stop,vaa3stop, vaa34stop, and v�aa34-157 were reduced by 69%, 31%,45%, and 38%, respectively, compared with the plaque sizes ofparental vP-Oka or the revertant viruses. To determine whetherORFS/L expression is abrogated in cells infected with vaa1stop,vaa3stop, or vaa34stop, we inserted an HA tag at the C termi-nus of ORFS/L, analogous to vS/L_cHA. Western blottingdemonstrated that ORFS/L expression was absent in cells in-fected with vaa34stop-cHA. Unexpectedly, however, bothvaa1stop-cHA and vaa3stop-cHA still expressed the ORFS/Lprotein, suggesting that the second start codon (aa 29) is usedfor the initiation of translation. The replication defect seenwith vaa34stop confirmed that the ORFS/L gene product isimportant for efficient VZV replication. It was apparent thatthe insertion of stop codons at aa1 or aa3 also resulted in a

FIG. 3. Localization of ORFS/L in infected and transfected cells.(A) Localization of ORFS/L (green) is shown, along with GM130(red), a marker for the Golgi apparatus, and DAPI (blue), for singleinfected MeWo cells (upper panels) or in syncytia (middle panels). Thelower panels show a focus on the nuclei within a syncytium that showscolocalization of ORFS/L protein with the nuclear envelope. Scalebars correspond to 30 �m (upper panels) and 100 �m (middle panels).(B) MeWo cells were transfected with expression plasmid pORFS/L-His and analyzed 48 h after transfection. ORFS/L (green) and GM130(red) were detected with the indicated antibodies. The transfected cellsare representative of the cell population. Scale bar corresponds to20 �m.

FIG. 4. Growth properties of viruses lacking the entire ORFS/L.(A) Plaque size measurements of v�aa1-157 (clone 1 [c1] and c2),vP-Oka, and revertant v�aa1-157-rev generated from clone 1. Recom-binant viruses were reconstituted in MeWo cells, IIF was performed 7days after transfection, and images were recorded. Twenty individualplaque areas were measured and are shown as averages (vP-Okaplaques were set at 100%) with standard deviations (error bars). *,Plaque areas induced by �aa1-157 mutants were significantly reduced(P � 0.0001) compared to those of parental vP-Oka or the revertantvirus by Student’s t test. (B) Representative plaque images for v�aa1-157 (clones 1 and 2), vP-Oka, and v�aa1-157-rev. Scale bars corre-spond to 200 �m.

VOL. 84, 2010 VZV ORFS/L (ORF0) AND REPLICATION 11665

growth defect which could not be attributed to absence of theORFS/L product. We concluded from the results that the se-vere replication defect seen in v�aa1-157 and vaa1stop and theless pronounced replication defect in vaa3stop were caused bya sequence element within the N-terminal 33 aa of the pre-dicted ORFS/L.

Cleavage defect of vaa1stop contributes to its growth defect.To elucidate whether the replication defect of vaa1stop andvaa3stop (Fig. 5A and B) was caused by an alteration of a

structural element present in the extreme 5� end of the pre-dicted ORFS/L, we addressed the cleavage of VZV genomesfrom concatemeric replication intermediates in cells infectedwith mutant or parental viruses. We infected MeWo cells withvaa1stop, vaa1stop-rev, vaa3stop, vaa34stop, or vP-Oka, har-vested the cells at 90 to 100% cytopathic effect, preparednuclear or whole-cell DNA, digested it with BamHI, andprobed for the genomic termini (Fig. 6A). While vaa1stop-revviruses showed a cleavage pattern comparable to that of theparental vP-Oka, viral DNA from cells infected with vaa1stopshowed an increase of uncleaved genomic termini in bothwhole-cell (Fig. 6B) and nuclear (Fig. 6C) VZV DNA. Nocleavage defect was detected with vaa34stop, suggesting thatthe elevated levels of uncleaved VZV genomic termini invaa1stop are not caused by reduced viral replication (Fig. 6C).Cleavage was only slightly, albeit discernibly, reduced in cellsinfected with vaa3stop, indicating that the two base pairs at thisposition are not as important as the two nucleotides changed invaa1stop for the generation of unit-length genomes in infectedcells (Fig. 6C). The relative quantity of uncleaved vaa1stopgenomes was significantly increased, by more than 3.6-fold inwhole-cellular and 2.7-fold in nuclear DNA, compared to thequantities of vP-Oka and vaa1stop-rev (Fig. 6D and E). Takentogether, we concluded from the results of the experimentsthat the extreme 5� region of the predicted ORFS/L is involvedin cleavage of unit-length genomes from replicative-form con-catemeric DNA before they are packaged into newly formednucleocapsids. This reduction in DNA cleavage could subse-quently result in impaired packaging of VZV genomes andcorrelates well with the growth defects of vaa1stop andvaa3stop. Taken together, the results indicate that the extreme5� terminus of ORFS/L contains a sequence element importantfor virus replication that is likely involved in the cleavage andpackaging of viral DNA.

DISCUSSION

Although the molecular tools available to study VZV repli-cation in vitro and in vivo have improved considerably over thepast decade, the functions of several of the 70 unique VZVORFs still remain unknown. While sequence similarities be-tween herpesviruses ORFs can often provide valuable infor-mation on protein localization and function, such predictionsonly provide testable hypotheses, not formal proof of the rolethese proteins or genetic elements play in related viruses.

For example, the function of ORFS/L, an ORF located atthe very end of UL, has remained enigmatic. Previously, it wasdemonstrated that a deletion of the C-terminal portion ofORFS/L resulted in a viral growth defect in vitro and in vivo. Ithas remained unclear, however, whether this defect is due toORFS/L gene function or a critical sequence element that isindependent of protein function per se. A second study alsosuggested that the ORFS/L protein localizes to the cytoplasm.Here, we chose to revisit both ORFS/L function and the lo-calization of the encoded protein and also to dissect putativelyoverlapping functions of this genomic region. We surmisedthat important functions are attributable to nucleotide se-quences specified in this region, which has close proximity tothe DNA processing elements involved in cleavage and pack-aging of the viral DNA into newly formed nucleocapsids.

FIG. 5. Growth properties of a panel of ORFS/L mutants.(A) Multistep growth kinetics of indicated viruses. Virus replication ofvaa1stop, vaa34stop, and v�aa34-157 was significantly reduced (P �0.05) at the indicated time points (asterisks) compared to that ofvP-Oka or the corresponding revertant virus as determined by Stu-dent’s t test. (B) Plaque size measurements of indicated viruses. *,Plaque areas of vaa1stop, vaa3stop, vaa34stop, and v�aa34-157 weresignificantly reduced (P � 0.001) compared to those of vP-Oka (set to100%) or the corresponding revertant virus as determined by Student’st test. Shown are the mean areas and standard deviations (error bars)for 60 individual plaques per virus. (C) Western blot analysis of cellsinfected with vP-Oka, vS/L_cHA, vaa1stop-cHA, vaa3stop-cHA, orvaa34stop-cHA. Samples were separated by SDS–15% PAGE andanalyzed by Western blotting with anti-HA antibodies.

11666 KAUFER ET AL. J. VIROL.

We first addressed the subcellular distribution of theORFS/L protein and were able to demonstrate that it localizesto the Golgi apparatus in infected and transfected cells, whichsupports previous bioinformatic predictions that suggested it isa homologue of the HSV-1 and HSV-2 UL56 gene products.Besides its subcellular localization, the position of ORFS/Lwithin the VZV genome, in close proximity to ORF3 thatrepresents the VZV homologue of UL55, also supports thisassumption (7). In addition, the ORFS/L protein shares 20%

sequence identity with HSV-1 and -2 and contains two of thethree conserved PY motives (L/PPXY). The PY motifs ofHSV-2 UL56 were shown to facilitate binding to Nedd4, an E3ubiquitin ligase involved in protein sorting and trafficking (4,26, 27). Although we show here that the subcellular localiza-tion of the ORFS/L protein is consistent with an involvementof Nedd4-dependent sorting, further studies will be necessaryto analyze the exact molecular functions of the ORFS/L prod-uct and its interaction with cellular proteins. The putative

FIG. 6. Cleavage of the viral genome in cells infected with mutant virus. (A) Schematic representation of the terminal fragments, TRS and UL,generated after restriction enzyme digestion with BamHI. The probe used to detect the genomic termini by Southern blotting is indicated. Thescale bar provides an accurate measure of the sizes of the uncleaved (2.8 kbp), the TRS (1.9 kbp), and the TRL (0.9 kbp) fragments expected aftercleavage and of the probe. (B) Cleavage analysis of whole-cell DNA derived from cells infected with vP-Oka, vaa1stop, or vaa1stop-rev. The P-OkaBAC (pP-Oka) served as a standard for uncleaved termini. The Southern blot shown is representative of the results of three independentexperiments. (C) Cleavage analysis of nuclear DNA derived from cells infected with vP-Oka, vaa1stop, vaa1stop-rev, vaa3stop, or vaa34stop. TheP-Oka BAC (pP-Oka) served as a standard for uncleaved termini. The Southern blot shown is representative of two independent experiments.(D) Quantification of band intensities of three independent cleavage assays using whole-cell DNA. The ratio of uncleaved to cleaved TRL DNAfragments is shown relative to the results for vP-Oka. *, The relative amount of uncleaved termini of vaa1stop was significantly increased (P � 0.05)compared to the results for vP-Oka and vaa1stop-rev as determined by Student’s t test. (E) Quantification of band intensities of two independentcleavage assays using nuclear DNA. The ratio of uncleaved to cleaved TRL DNA fragments is shown relative to the results for vP-Oka.

VOL. 84, 2010 VZV ORFS/L (ORF0) AND REPLICATION 11667

interaction partners include KIF1A, a member of the kinesinfamily that is involved in vesicular trafficking along microtu-bules and for which an interaction with the HSV-2 UL56 prod-uct was identified (17).

Western blot analysis showed that the P-Oka ORFS/L isexpressed as a single protein species, and the apparent molec-ular mass of 18 kDa, as determined by SDS-PAGE and West-ern blotting, corresponded nicely to the predicted molecularmass of the full-length HA-tagged protein (Mr � 18,207).However, the data accumulated here indicate that translationis initiated from the 2nd and not the 1st start codon of thepredicted ORFS/L. This finding is in agreement with the re-sults of a recent study that demonstrated that plasmid-basedtransient expression of P-Oka ORFS/L initiated from the 2ndpredicted start codon resulted in a protein with an apparentmolecular mass of approximately 17 kDa (18), which is 3 kDalarger than predicted. This report also confirmed that the dis-crepancy between the predicted and apparent molecular massdetermined here for the P-Oka ORFS/L protein is not due toglycosylation (18). Similarly, in silico predictions indicate thatORFS/L is not myristoylated, which could also account for adeviation of the apparent from the predicted molecular mass.The lack of a signal peptide indicates that ORFS/L is likely nottranslated at the rough endoplasmic reticulum (ER) or modi-fied in the ER or Golgi network. Similar to the HSV-1 andHSV-2 UL56 proteins, the VZV ORFS/L product is predictedto encode a C-terminally anchored, type II transmembraneprotein. Type II transmembrane proteins do not posses N-terminal signal sequences that would provide a signal for mem-brane insertion. Instead, they contain a C-terminal hydropho-bic region facilitating insertion and orienting the N terminus ofthe protein toward the cytoplasm (19). In the case of theHSV-2 UL56 protein, the C-terminal hydrophobic region wasindeed shown to be responsible for membrane insertion and itslocalization to the Golgi apparatus (16, 17).

It is notable that, in contrast to the 17 wild-type VZVstrain sequences available in GenBank (http://www.ncbi.nlm.nih.gov/), the vaccine strains V-Oka, VariVax, and VarilRixharbor a mutation in the stop codon that extends the ORF by276 bp (92 aa). In contrast to the ORFS/L product of P-Oka,that of V-Oka was shown to be glycosylated (18), but theextended C-terminal domain in V-Oka still allows its insertioninto membranes and incorporation into the virion (18).

Mutations of ORFS/L immediately downstream of the de-termined start codon confirmed that the predicted ORFS/Lgene product plays an important albeit not essential role inVZV replication. The involvement in vesicular trafficking, asshown for HSV-2 UL56, could account for the growth defectobserved in cultured MeWo and HFF cells. In contrast, partialand complete deletions of ORFS/L sequences revealed thatthe 5� region of ORFS/L is essential for replication. Analysis ofthe status of viral DNA in cells infected with the vaa1stop andvaa3stop mutant viruses showed that a sequence element con-tained in the predicted ORFS/L is involved in the cleavage and,consequently, packaging of single-unit genomes from concate-meric replication intermediates. In our experiments, compara-ble amounts of cleaved TRS and TRL fragments were presentin vaa1stop-infected cells and cells infected with the parentalvP-Oka. This apparent conundrum is likely caused by the ex-tended time of about 96 h, until complete cytopathic effect was

reached, allowed for infection in the case of vaa1stop. In someexperiments, we also detected a minimal size difference in thevaa1stop TRL fragment, which could potentially be caused byterminal cleavage at noncanonical sequences.

Intriguingly, the sequence that we showed is necessary forefficient cleavage of viral DNA is not present at the oppositeend of the UL, which could result in preferential packaging ofviral DNA with only one orientation of the UL. The asymmetryof the ends of the VZV UL region was first detected anddiscussed by Davison and coworkers, who predicted a cleavageelement within the UL region that resides at least 90 bp fromthe genomic terminus, which exactly correlates with the 5�region of ORFS/L (6, 7). Whether a similar cleavage elementat the beginning of the UL region is responsible for the direc-tional packaging of other members of the Varicellovirus genusstill remains a possibility.

In conclusion, our study provides further evidence that theORFS/L gene product plays a role in VZV replication and thatthe 5� region is involved in cleavage and packaging. Furtherstudies of ORFS/L function and its potential interaction part-ners will shed light on its detailed role in VZV replication. Inaddition, defining the essential sequence elements within the 5�region of ORFS/L will increase the understanding of its role incleavage and packaging, as well as the almost exclusive orien-tation of the UL in VZV virions.

ACKNOWLEDGMENTS

This study was supported by PHS grant 1R21AI061412 and anunrestricted grant from the Freie Universitat Berlin to N.O.

REFERENCES

1. Arvin, A. M. 1996. Varicella-zoster virus. Clin. Microbiol. Rev. 9:361–381.2. Arvin, A. M. 2001. Varicella-zoster virus, p. 2731–2767. In B. N. Fields, D. M.

Knipe, and P. M. Howley (ed.), Virology. Lippincott-Raven Publishers, Phil-adelphia, PA.

3. Berkowitz, C., M. Moyal, A. Rosen-Wolff, G. Darai, and Y. Becker. 1994.Herpes simplex virus type 1 (HSV-1) UL56 gene is involved in viral intra-peritoneal pathogenicity to immunocompetent mice. Arch. Virol. 134:73–83.

4. Bruce, M. C., V. Kanelis, F. Fouladkou, A. Debonneville, O. Staub, and D.Rotin. 2008. Regulation of Nedd4-2 self-ubiquitination and stability by a PYmotif located within its HECT-domain. Biochem. J. 415:155–163.

5. Carey, T. E., T. Takahashi, L. A. Resnick, H. F. Oettgen, and L. J. Old. 1976.Cell surface antigens of human malignant melanoma: mixed hemadsorptionassays for humoral immunity to cultured autologous melanoma cells. Proc.Natl. Acad. Sci. U. S. A. 73:3278–3282.

6. Davison, A. J. 1984. Structure of the genome termini of varicella-zoster virus.J. Gen. Virol. 65(Pt. 11):1969–1977.

7. Davison, A. J. 2007. Comparative analysis of the genomes, p. 10–26. In A.Arvin, G. Campadielli-Fiume, E. Mocarski, P. S. Moore, B. Roizman, B.Kramarsky, S. Z. Salahuddin, and R. C. Gallo (ed.), Human herpesviruses:biology, therapy and immunoprophylaxis. Cambridge University Press, Cam-bridge, United Kingdom.

8. Davison, A. J. 2010. Herpesvirus systematics. Vet. Microbiol. 143:52–69.9. Davison, A. J., and N. M. Wilkie. 1983. Location and orientation of homol-

ogous sequences in the genomes of five herpesviruses. J. Gen. Virol. 64(Pt.9):1927–1942.

10. Ecker, J. R., and R. W. Hyman. 1982. Varicella zoster virus DNA exists astwo isomers. Proc. Natl. Acad. Sci. U. S. A. 79:156–160.

11. Gilden, D. H., R. Gesser, J. Smith, M. Wellish, J. J. Laguardia, R. J. Cohrs,and R. Mahalingam. 2001. Presence of VZV and HSV-1 DNA in humannodose and celiac ganglia. Virus Genes 23:145–147.

12. Hammerschmidt, W., H. Ludwig, and H. J. Buhk. 1988. Specificity of cleav-age in replicative-form DNA of bovine herpesvirus 1. J. Virol. 62:1355–1363.

13. Kemble, G. W., P. Annunziato, O. Lungu, R. E. Winter, T. A. Cha, S. J.Silverstein, and R. R. Spaete. 2000. Open reading frame S/L of varicella-zoster virus encodes a cytoplasmic protein expressed in infected cells. J. Vi-rol. 74:11311–11321.

14. Kinchington, P. R., W. C. Reinhold, T. A. Casey, S. E. Straus, J. Hay, andW. T. Ruyechan. 1985. Inversion and circularization of the varicella-zostervirus genome. J. Virol. 56:194–200.

15. Klupp, B. G., C. J. Hengartner, T. C. Mettenleiter, and L. W. Enquist. 2004.

11668 KAUFER ET AL. J. VIROL.

Complete, annotated sequence of the pseudorabies virus genome. J. Virol.78:424–440.

16. Koshizuka, T., F. Goshima, H. Takakuwa, N. Nozawa, T. Daikoku, O. Koi-wai, and Y. Nishiyama. 2002. Identification and characterization of the UL56gene product of herpes simplex virus type 2. J. Virol. 76:6718–6728.

17. Koshizuka, T., Y. Kawaguchi, and Y. Nishiyama. 2005. Herpes simplex virustype 2 membrane protein UL56 associates with the kinesin motor proteinKIF1A. J. Gen. Virol. 86:527–533.

18. Koshizuka, T., M. Ota, K. Yamanishi, and Y. Mori. 2010. Characterizationof varicella-zoster virus-encoded ORF0 gene—comparison of parental andvaccine strains. Virology 405:280–288.

19. Kutay, U., E. Hartmann, and T. A. Rapoport. 1993. A class of membraneproteins with a C-terminal anchor. Trends Cell Biol. 3:72–75.

20. McVoy, M. A., D. E. Nixon, J. K. Hur, and S. P. Adler. 2000. The ends onherpesvirus DNA replicative concatemers contain pac2 cis cleavage/packag-ing elements and their formation is controlled by terminal cis sequences.J. Virol. 74:1587–1592.

21. Rosen-Wolff, A., and G. Darai. 1991. Identification and mapping of the UL56gene transcript of herpes simplex virus type 1. Virus Res. 19:115–126.

22. Sambrook, J., D. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: alaboratory manual. Cold Spring Harbor Laboratory Press, Cold Spring Har-bor, NY.

23. Slobedman, B., and A. Simmons. 1997. Concatemeric intermediates ofequine herpesvirus type 1 DNA replication contain frequent inversions ofadjacent long segments of the viral genome. Virology 229:415–420.

24. Tischer, B. K., B. B. Kaufer, M. Sommer, F. Wussow, A. M. Arvin, and N.Osterrieder. 2007. A self-excisable infectious bacterial artificial chromosomeclone of varicella-zoster virus allows analysis of the essential tegument pro-tein encoded by ORF9. J. Virol. 81:13200–13208.

25. Tischer, B. K., J. von Einem, B. Kaufer, and N. Osterrieder. 2006. Two-stepred-mediated recombination for versatile high-efficiency markerless DNAmanipulation in Escherichia coli. Biotechniques 40:191–197.

26. Ushijima, Y., F. Goshima, H. Kimura, and Y. Nishiyama. 2009. Herpessimplex virus type 2 tegument protein UL56 relocalizes ubiquitin ligaseNedd4 and has a role in transport and/or release of virions. Virol. J.6:168.

27. Ushijima, Y., T. Koshizuka, F. Goshima, H. Kimura, and Y. Nishiyama.2008. Herpes simplex virus type 2 UL56 interacts with the ubiquitin ligaseNedd4 and increases its ubiquitination. J. Virol. 82:5220–5233.

28. Zhang, Z., J. Rowe, W. Wang, M. Sommer, A. Arvin, J. Moffat, and H. Zhu.2007. Genetic analysis of varicella-zoster virus ORF0 to ORF4 by use of anovel luciferase bacterial artificial chromosome system. J. Virol. 81:9024–9033.

VOL. 84, 2010 VZV ORFS/L (ORF0) AND REPLICATION 11669