JOURNAL OF VIROLOGY, Sept. 1998, p. 6997–7004 0022-538X/98/$04.0010 Copyright © 1998, American Society for Microbiology. All Rights Reserved.

Vol. 72, No. 9

Identification of Persistent RNA-DNA Hybrid Structures within the Origin of Replication of Human Cytomegalovirus MARK N. PRICHARD,1 SANJU JAIRATH,2 MARK E. T. PENFOLD,3 STEPHEN ST. JEOR,4 MARLENE C. BOHLMAN,4 AND GREGORY S. PARI4* Iconix Pharmaceuticals, Inc.,1 and Aviron, Inc.,3 Mountain View, California 94043; Hybridon, Inc., Cambridge, Massachusetts 021392; and Department of Microbiology, School of Medicine, University of Nevada—Reno, Reno, Nevada 895574 Received 11 February 1998/Accepted 22 May 1998

Human cytomegalovirus (HCMV) lytic-phase DNA replication initiates at the cis-acting origin of replication, oriLyt. oriLyt is a structurally complex region containing repeat elements and transcription factor binding sites. We identified two site-specific alkali-labile regions within oriLyt which flank an alkali-resistant DNA segment. These alkali-sensitive regions were the result of the degradation of two RNA species embedded within oriLyt and covalently linked to viral DNA. The virus-associated RNA, vRNA, was identified by DNase I treatment of HCMV DNA obtained from sucrose gradient purified virus. This heterogeneous population of vRNA was end labeled and used as a hybridization probe to map the exact location of vRNAs within oriLyt. vRNA-1 is localized between restriction endonuclease sites XhoI at nucleotide (nt) 93799 and SacI at nt 94631 and is approximately 500 bases long. The second vRNA, vRNA-2, lies within a region which exhibits a heterogeneous restriction pattern located between the SphI (nt 92636) and BamHI (nt 93513) and is approximately 300 bases long. This region was previously shown to be required for oriLyt replication (D. G. Anders, M. A. Kacica, G. S. Pari, and S. M. Punturieri, J. Virol. 66:3373–3384, 1992). RNase H analysis determined that vRNA-2 forms a persistent RNA-DNA hybrid structure in the context of the viral genome and in an oriLytcontaining plasmid used in the transient-replication assay. Origin-dependent replication of HCMV in primary human cells requires the gene products of the core replication open reading frames in addition to UL36–38, UL112–113, IE1/IE2 and UL84 (22, 27). In Vero cells, however, UL84 appears to be the only non-core replication protein required for oriLyt-dependent replication (27). These HCMV characteristics, i.e., (i) the complex structure of oriLyt, (ii) the presence of an intra-origin RNA transcript, and (iii) the lack of an apparent trans-acting factor functioning as an initiator, suggest the possibility of a unique method of DNA replication for HCMV. In an effort to define a mechanism of initiation of DNA replication, we examined DNA structure within oriLyt and identified the presence of alkali-labile regions. The existence of alkali-sensitive regions suggests the presence of abasic sites or RNA, which is possibly embedded within oriLyt. As a consequence of the identification of NaOH-sensitive regions within oriLyt, we developed a method for extracting the RNA component associated with packaged virion DNA (vRNAs). In this report, we show that two vRNAs, approximately 300 and 500 bases long, are present as two stable and persistent RNA-DNA hybrid regions within oriLyt. Transient-replication experiments show that vRNAs are incorporated within oriLyt at the time of, or after, DNA replication and are dependent upon DNA synthesis. One of these RNA-DNA hybrid regions maps to a previously identified segment of oriLyt that is indispensable for origin function (1, 18). In addition, we have identified the presence of variably repeated sequence motifs within the HCMV origin, corresponding to one of these RNA-DNA hybrid regions.

Initial reports describing the identification and characterization of the human cytomegalovirus (HCMV) origin of replication, oriLyt, have demonstrated that this replication origin bears little resemblance to other previously identified origins within the herpesvirus family (1, 2, 7, 8, 10–12, 17, 28, 31, 32). The HCMV core origin spans approximately a 3-kb region from nucleotide (nt) coordinates 91321 to 93715. Other flanking sequences which augment origin function have also been defined (1, 12, 18). Within the core origin, multiple consensus transcription factor binding sites and novel repeat elements have been identified. However their role in replication has not been determined (1). Recently, Huang et al. identified an RNA transcript approximately 200 bases long, originating within oriLyt (14). This observation suggests the possibility that HCMV has a mode of initiation of DNA replication unlike that of other herpesviruses. This RNA species, SRT (smallest replicator transcript), was detected in the presence of phosphonoformic acid and mapped to regions within the origin shown to be required for oriLyt-dependent DNA replication (14). In addition to various cis-acting elements within HCMV oriLyt, another interesting feature is the apparent lack of a replication initiator protein analogous to the herpes simplex virus type 1 origin binding protein, UL9 (22, 23). In HSV-1, this protein binds to the origin of replication and initiates DNA synthesis by an apparent helicase activity (3, 9, 15, 21, 33). Replication at the Epstein-Barr virus oriLyt also appears to be facilitated by the initiator protein, Zta (10). Zta is a transactivator that binds to sites within the Epstein-Barr virus oriLyt and is postulated to initiate DNA replication by either the activation of transcription or protein-protein interactions (10).

MATERIALS AND METHODS

* Corresponding author. Mailing address: Dept. of Microbiology, School of Medicine, University of Nevada—Reno, Reno, NV 89557. Phone: (702) 784-1383.

Cells and virus. Human foreskin fibroblasts were used for all experiments and propagated in Dulbecco’s modified Eagle’s medium supplemented with 10%

6997

6998

PRICHARD ET AL.

FIG. 1. Identification of site-specific alkali-sensitive regions within HCMV oriLyt. Viral DNA isolated from sucrose gradient centrifugation was cleaved with NheI and incubated at 70°C in 50 mM NaOH or 50 mM NaCl. Samples were neutralized with Tris-HCl (pH 7.0), separated on a 5% denaturing polyacrylamide gel, transferred to a nylon membrane, and hybridized with pOri. Lanes: 1, AD169 DNA treated with NaCl; 2, AD169 DNA treated with NaOH. The arrow indicates the presence of a 550-base band on NaOH-resistant DNA.

(vol/vol) fetal calf serum. HCMV AD169 (ATCC V-538) was propagated as described previously (29). Isolation of vRNA. Human foreskin fibroblasts infected with HCMV (multiplicity of infection, 1 PFU/cell) from 10 roller bottle cultures (850 cm2) were harvested 15 days postinfection by using a cell scraper. Infected cells were transferred to 250-ml centrifuge bottles. Infected cells were sonicated by using 5 to 10 pulses with a microtip probe to release intracellular virus. Cell debris was removed by low-speed centrifugation, and the supernatant was layered over a 35% (wt/vol) sucrose cushion in a 38-ml SW28 ultracentrifuge tube. Virus particles were pelleted (110,000 3 g) for 1 h at 4°C. The resulting viral pellet was resuspended in Tris-EDTA (TE) and centrifuged through a 30 to 60% continuous sucrose gradient by ultracentrifugation (110,000 3 g) for 1 h at 4°C. The virus band was removed, and the sucrose-virus mixture was diluted with TE. Virions were again pelleted by ultracentrifugation (110,000 3 g) for 1 h, and these final virus pellets were resuspended in 300 ml of TE. A 100-ml volume of virus was treated with 5 U of RNase A for 30 min (final volume, 200 ml) at 37°C in RNase A buffer (100 mM Tris-HCl [pH 7.4], 150 mM NaCl, 10 mM MgCl2, 1 mM dithiothreitol). After RNase A incubation, proteinase K, sodium dodecyl sulfate (SDS), and EDTA (final concentrations, 1 mg/ml, 1% [wt/vol], and 10 mM, respectively) were added. This mixture was incubated for 3 h at 60°C. After incubation, sodium acetate (pH 5.5) was added to a final concentration of 0.3 M and the solution was extracted twice with phenol-chloroform-isoamyl alcohol (25:24:1) and then twice with chloroform-isoamyl alcohol (24:1). The aqueous-phase solution was removed, and viral DNA was precipitated with 2 volumes of 100% ethanol. Viral DNA was resuspended in TE and adjusted to a final concentration of 1 mg/ml. A 50-mg sample of HCMV DNA was heated for 5 min at 95°C and immediately placed on ice for 3 min. Then 50 U of RNase-free DNase I was added by using DNase I buffer (100 mM Tris-HCl [pH 7.4], 150 mM NaCl, 10 mM MgCl2, 1 mM dithiothreitol), and the mixture was incubated for 1.5 h at 37°C. vRNA was isolated with Trizol reagent (Bethesda Research Laboratories) as specified by the manufacturer. The vRNA pellet was resuspended in 50 ml of TE. DNA constructs. The oriLyt-containing plasmid, pOri, was constructed by subcloning the KpnI (nt 89797)-PvuII (nt 94860) fragment directly from the AD169 genome into the KpnI-EcoRV site of pGEM7zf(2) (Promega, Madison, Wis.). pOri XhoA and pOri XhoB were made by cleaving pOri with XhoI and ligating the resulting 1.1- and 1.7-kb fragments, respectively, into XhoI-cleaved pBlueScript KS(2) (Stratagene, La Jolla, Calif.). pOri BX was constructed by cleaving pOri XhoB with XhoI and then subjecting it to partial cleavage with BamHI, excising the 300-bp fragment, and ligating it into BamHI-XhoI-cleaved pGEM7zf(2). Construction of plasmids containing HCMV replication genes required for oriLyt-dependent DNA replication was described previously (22). The oriLyt subclone used to generate riboprobes was constructed by cleaving pOri with SphI (nt 93513) and BamHI (nt 94636) and ligating the resulting fragment into SphI-BamHI-cleaved pGEM7zf(2). pOri XS was constructed by cleaving pOri SB-A with XhoI and SacI and ligating the resulting 830-bp fragment into XhoI-SacI-cleaved pBlueScript SK(2). The subclone pOri-HT, which contains the heterogeneous region within oriLyt, was made by PCR amplification of AD169 DNA with the primers 59-ATG GAAAACCTATATATAAGGAGGGGT-39 and 59-CTGGGTGGGGGATCC CCGGTCGCCCAC-39. The resulting PCR product was ligated into pCRII

J. VIROL. (Invitrogen) and sequenced with T7 and SP6 sequencing primers and internal primers when necessary. NaOH treatment of HCMV DNA. A 5-mg portion of HCMV DNA AD169 was cleaved with NheI and subsequently treated with 50 mM NaOH or 50 mM NaCl, heated to 65°C for 15 min, and neutralized with Tris-HCl (pH 7). Single-stranded fragments were separated through a 5% denaturing polyacrylamide gel (5 M urea) and transferred to a nylon membrane. The membrane was hybridized with 32 P-random-primer-labeled pOri plasmid. Mapping of vRNA to HCMV oriLyt: Southern blot analysis. Plasmid pOri was cleaved with various restriction enzymes as described in the figure legends, separated on a 1% agarose gel, and transferred to ZetaProbe (Bio-Rad) nylon membrane. vRNA 32P-end-labeled probes were generated by using 10 U of T4 polynucleotide kinase (New England BioLabs), 20 mCi of [g-32P]ATP, and approximately 5 mg of vRNA. The membranes were incubated in a Robbins Scientific hybridization oven for 14 h at 65°C with labeled vRNA by using 5 ml of hybridization buffer (1.53 SSPE, 7% SDS, 10% [wt/vol] polyethylene glycol 4000). The blots were washed twice for 15 min each at room temperature with 23 SSC (13 SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–1% SDS and then twice with 0.13 SSC–1% SDS for 45 min each at 65°C. The blots were then exposed to BioMax X-ray film (Kodak) for 15 h at 280°C. Northern analysis. vRNA (20 to 50 mg) was denatured for 15 min at 70°C in RNA sample buffer (50% formamide, 5% formaldehyde, 13 morpholinepropanesulfonic acid [MOPS]), separated on a 3% formaldehyde–agarose gel (3:1 Nusieve [FMC Corp.] agarose), and transferred to ZetaProbe Nylon membrane. Membranes were hybridized with random-primer-generated 32P-labeled plasmid pOri in hybridization buffer for 14 h at 60°C and washed as described above. For RNA probes, hybridization was performed with riboprobe buffer (1.53 SSPE, 1% SDS, 50% formamide, 100 mg of salmon sperm DNA per ml, 20 mg of tRNA per ml). Probe concentrations were adjusted accordingly so that the same specific activity was used for each riboprobe representing each RNA strand. The membranes were incubated for 14 h at 65°C. RNase H treatment of HCMV DNA. Sucrose gradient-purified virus DNA obtained as described above was treated with 10 U of Escherichia coli RNase H (Boehringer Mannheim) for 1 h at 37°C. The DNA-RNase H mixture was extracted once with phenol-chloroform-isoamyl alcohol (25:24:1) and once with chloroform-isoamyl alcohol (24:1), ethanol precipitated, and resuspended in TE. DNA samples were then cleaved with BglII and PvuII. Single-stranded DNA fragments were separated on a 1% formaldehyde–agarose gel and transferred to a nylon membrane. The blots were hybridized with 32P-random-primer-labeled pOri SB-B. The transient-replication assay was performed as described previously (23). Total cellular DNA was extracted with phenol-chloroform, cleaved with HindIII,

FIG. 2. Two species of vRNAs exist within oriLyt. vRNA was isolated from sucrose gradient-purified AD169 DNA as described in Materials and Methods. vRNA was separated on a formaldehyde-agarose gel, transferred to a nylon membrane, and probed with pOri. Lanes: 1, vRNA; 2, vRNA treated with RNase A prior to gel loading. The arrows indicate the presence of a RNA band corresponding to a 500 bases and an RNA species of approximately 300 bases.

VOL. 72, 1998

RNA-DNA HYBRID AT oriLyt

6999

FIG. 3. Linear map of HCMV oriLyt. The relative positions and nucleotide sequence coordinates of certain restriction endonuclease sites are shown at the top. Six oriLyt subclones used for mapping experiments are shown below the map.

and treated with RNase H. DNA was then treated in the same as the viral DNA described above. Duplicate samples were treated with HindIII and DpnI to ensure that replicated plasmid DNA was present in samples that included all the required genes in the cotransfection mixture.

RESULTS Identification of site-specific alkali-labile sites within HCMV oriLyt. Viral DNA from the HCMV (AD169) genome was subjected to mild-alkali hydrolysis in an attempt to detect the presence of abasic sites (i.e., nucleotides in which the base component is missing) in virion DNA. Unmodified DNA is unaffected under these conditions, but abasic sites or ribonucleotides will be degraded and the cleavage products can be detected on denaturing gels. Sucrose gradient-purified virus was isolated and HCMV DNA was extracted as described in Materials and Methods. HCMV DNA was cleaved with NheI and then treated with either NaOH or NaCl. Single-stranded fragments were separated through a denaturing gel and probed with oriLyt sequences from nt 89797 to 94860 (pOri). Figure 1 is an autoradiogram of a Southern blot of HCMV DNA treated with either NaOH or NaCl. The arrow indicates a DNA band corresponding to a fragment of approximately 550 bases in the alkalitreated sample which was not detected in the NaCl-treated sample (Fig. 1). This result indicated that there were alkalilabile sites within oriLyt of HCMV. HCMV oriLyt contains two species of vRNA. The presence of NaOH-sensitive regions within oriLyt prompted us to investigate if RNA, instead of DNA or regions containing abasic sites, could be integrated or embedded within the HCMV genome. To this end we developed a method to isolate possible RNA components associated with packaged viral DNA. After the isolation of sucrose gradient-purified virus, virus particles

were treated with RNase A to degrade exogenous RNA which could have copurified with the virus. Viral protein components were eliminated by incubation of virus with proteinase K and SDS followed by phenol-chloroform extraction. HCMV genomic DNA was isolated by ethanol precipitation and resuspended in TE. Viral DNA was heated to 95°C to denature the DNA strands and subjected to treatment with RNase-free DNase I. This effectively removed all of the DNA components of the viral genome. Following DNase I treatment, an RNA isolation procedure was performed. This resultant RNA was termed vRNA, defining the species associated with viral DNA. Figure 2 is an autoradiogram of a Northern blot in which total vRNA was hybridized with the plasmid containing cloned oriLyt, pOri. The arrows indicate that two RNA species were detected (Fig. 2, lane 1). As a control, total vRNA was treated with RNase A before being loaded onto the gel and probed along with untreated samples (lane 2). Two vRNAs with approximate sizes of 300 and 500 bases were detected. This data indicated that RNA was associated with the HCMV genome, was complementary to oriLyt sequences as demonstrated by Northern blot hybridization, and was RNase A sensitive. vRNAs are localized within two regions of oriLyt. After the identification of vRNAs within oriLyt, we wanted to define their approximate location within the origin. pOri was cleaved with various restriction endonucleases and subjected to gel electrophoresis. Southern blots of the resulting DNA fragments were probed with 32P-end-labeled vRNA. A schematic representation of the cloned HCMV oriLyt fragment, pOri, which was used in restriction endonuclease mapping experiments is shown at the top of Fig. 3. Below the pOri linear map are the coordinates of six pOri subclones also used in mapping studies. Figure 4 is an autoradiogram of a Southern blot of pOri

7000

PRICHARD ET AL.

FIG. 4. Mapping of HCMV oriLyt vRNAs. Shown is an autoradiogram of a Southern blot of pOri cleaved with various restriction endonucleases and probed with end-labeled vRNA. Lanes: 1, BamHI-NotI; 2, SphI-BamHI; 3, NotI-NsiI; 4, XhoI-NotI; 5, XhoI-EcoRI. Arrows indicate the hybridization of vRNA to the 4.3-kb BamHI fragment and the 900-bp BamHI-NotI fragment.

fragments probed with vRNA. pOri cleaved with BamHI and NotI produces four fragments, two of which hybridize with vRNA (Fig. 4, lane 1). One is the 4.3-kb fragment containing oriLyt sequences from the NotI (nt 92888) site to nt 94861 plus 3 kb of vector sequences. The other fragment is the NotI (nt 92888)-BamHI (nt 93513) 1.1-kp fragment (Fig. 3 and 4). This hybridization pattern reveals that vRNA is complementary to the region of oriLyt corresponding to nt 92888 to 94861 and that no hybridization signal was detected from nt 89796 to 92888 (Fig. 3). To confirm these hybridization results, a series of additional DNA hybridization experiments were performed. For example, the SphI-BamHI cleavage of pOri shows that the vRNA probe hybridized only to the 1.4-kb and 900-bp fragments (Fig. 4, lane 2) corresponding to nt 93361 to 94861 and nt 92866 to 93513, respectively (Fig. 3, linear map). In addition, cleavage with NotI and NsiI shows that no hybridization signal was detected between nt 92399 (NsiI) and 92888 (NotI). The 4.9-kb band results from the NotI (nt 92888)-NsiI (nt 94861) fragment plus 3 kb of vector sequence. No hybridization signal corresponding to vector sequences was detected. A faint 3-kb band was detected in Fig. 4, lane 2. Further experiments were performed to determine if this band was the result of a weak nonspecific hybridization signal or if vRNAs extend into this region. Results from hybridization studies performed on XhoI-NotI and XhoI-EcoRI cleavages also indicated that the two species of vRNA are contained entirely within the 2-kb region (nt 92888 to 94861) of oriLyt. vRNA hybridization signals resulting from the probing of Southern blots of an XhoI-NotI cleavage indicated that two bands were detected, corresponding to nt 92636 to 93799 and 93799 to 94861 (Fig. 4, lane 4). vRNA also hybridized to two of the fragments from an XhoI-EcoRI cleavage corresponding to nt 92888 to 93799 and 93799 to 94861 (lane 5). These results were the first indication that additional DNA sequence was present within the pOri clone. This extra DNA was not reported in the previously published HCMV sequence (5). Additional DNA sequence was also demonstrated when a XhoI-EcoRI cleavage pattern of pOri was examined. When

J. VIROL.

probed with vRNA, the predicted hybridization bands should yield XhoI fragments of 1.2 and 1.05 kb. However, the 1.05-kb fragment was detected as an approximately 1.4-kb DNA band (Fig. 4, lane 5). This was confirmed when a XhoI-NotI cleavage was performed. A 900-bp band corresponding to the NotI-XhoI fragment (nt 92888 to 93799) is predicted (5). However, the DNA band detected was approximately 1.2 kb (lane 4). Based on these observations, we conclude that our subclone of oriLyt, pOri, contained an extra 300 bp of DNA sequence between the NotI site at nt 92888 and the BamHI site at nt 93513. One species of vRNA maps to an essential region of oriLyt. To further define the location of HCMV vRNAs within oriLyt, we subcloned four segments of oriLyt corresponding to sequences from the NotI site at nt 92888 to nt 94861 (Fig. 3, bottom). Subclone pOri XhoA was cleaved with MluI and EcoRI, and Southern blots of the resulting DNA fragments were probed with vRNA. vRNA hybridized to two fragments of an MluIEcoRI cleavage, indicating that this species of vRNA spans the MluI site (Fig. 5A, lane 1, and Fig. 3, bottom). However, a HincII cleavage of pOri SB-A resulted in hybridization to the two HincII fragments and no other region, indicating that this vRNA species was localized between the flanking HincII sites (Fig. 5A, lane 4, and Fig. 3). This species of vRNA is referred to below as vRNA-1. The localization of the other species of vRNA was determined with the subclone pOri XhoB (Fig. 3, bottom). A BamHI cleavage of pOri XhoB produces three fragments. Only one of these was detected by the vRNA probe (Fig. 5A, lane 2). This 1.3-kb BamHI fragment corresponds to the BamHI fragment at the left end of the pOri XhoB clone (Fig. 3, bottom). This result indicated that this vRNA, subsequently called vRNA-2, was contained within nt 92636 to 93513 (Fig. 3). To further define the location of vRNA-2, pOri XhoB was cleaved with NotI and BamHI. The resulting fragments were probed with end-labeled vRNA. vRNA-2 hybridized to the NotI-BamHI fragment corresponding to nt 92888 to 93513, indicating that vRNA-2 maps completely within this fragment (Fig. 5A, lane 3, and Fig. 3, bottom). In conclusion, vRNA-1 is located between the XhoI (nt

FIG. 5. vRNA-2 maps within an essential region of oriLyt. Shown are subclones of pOri cleaved with various restriction endonucleases and probed with vRNA. (A) Lanes: 1, subclone pOri XhoA cleaved with MluI-EcoRI; 2, subclone pOri XhoB cleaved with BamHI; 3, subclone pOri XhoB cleaved with NotIBamHI; 4, subclone pOri SB-A cleaved with HincII. (B) Subclones of pOri indicating the locations of vRNA-1 and vRNA-2. Lanes: 1, pOri BX cleaved with XhoI-SacI; 2, pOri SB-B cleaved with SphI-BamHI; 3, pOri XS cleaved with XhoI-SacI.

VOL. 72, 1998

FIG. 6. Detection of vRNA-2 with strand-specific probes. Shown is an autoradiogram of a Northern blot where vRNA was separated on a formaldehyde gel and probed with strand-specific riboprobes. Lanes: 1, vRNA hybridized with a riboprobe generated in the nt 92576-to-93513 direction (top strand); 2, vRNA hybridized with a riboprobe generated in the nt 93513-to-92576 direction (bottom strand). The arrow in lane 2 indicates the broad band detected when probes generated from the top strand were used.

93799)-SacI (nt 94636) sites and vRNA-2 is contained entirely within the restriction fragment SphI (nt 92866)-BamHI (nt 93513), (Fig. 5B, lanes 2 and 3, respectively). Based on these results, we conclude that the hybridization signal seen in Fig. 4, lane 2, was the result of a weak nonspecific binding to the 3-kb fragment. No hybridization signal was detected when vRNA was used as a probe against sequences within the restriction fragment BamHI (nt 93513)-XhoI (nt 93799) (Fig. 5B, lane 1, and Fig. 3). The region between nt 92866 and 93513 was previously shown to be essential for oriLyt function (1). The location of vRNAs also explains the results observed in Fig. 1, where a 550-bp alkali-resistant fragment was detected in Southern blots. This fragment was the result of the degradation of the two vRNA segments which flanked an NaOH-insensitive 500-bp DNA region. vRNA-2 originates from the same DNA strand as SRT. Recently, a transcript of approximately 200 bases was identified within oriLyt (14). This transcript, SRT, is localized between nt 92431 and 92688 (14). The 59 end of SRT is located approximately 180 nt upstream of the 39 end of vRNA-2. To determine if there could be some relationship between vRNA-2 and SRT, we investigated if vRNA-2 originated from the same strand as SRT. To this end, we performed Northern blotting with singlestranded riboprobes specific for vRNA-2. Total vRNA was probed with riboprobes generated from plasmid pOri SB-B (Fig. 3). Two riboprobes were made, one generated from the bottom (nt 93513 to 92576) strand and the other from the top (nt 92576 to 93513) strand. When probes generated from the top strand were used, a broad band was detected in the autoradiogram of a Northern blot of vRNA (Fig. 6, lane 2). However, no distinct band was detected when riboprobes made from the bottom strand were used to probe vRNA (Fig. 6, lane 1). This indicated that vRNA-2 originated from the top strand, which is the same strand as the previously identified SRT (14).

RNA-DNA HYBRID AT oriLyt

7001

vRNA-2 is present as an RNA-DNA hybrid. vRNA was isolated from packaged viral DNA and was mapped to regions within oriLyt where NaOH-sensitive structures were shown to exist. To show that vRNA was integrated in the HCMV genome and to confirm that the alkali sensitivity was due to the presence of an RNA-DNA hybrid, we used the bacterial enzyme RNase H, an endoribonuclease which specifically degrades the RNA component of an RNA-DNA hybrid structure, to locate hybrid structures within oriLyt. Viral DNA was isolated as described in Materials and Methods, cleaved with BglII and PvuII, and treated with 10 U of RNase H. DNA fragments were separated as single-stranded DNA on a denaturing gel and hybridized with pOri SB-B. Figure 7A is an autoradiogram of a Southern blot of RNase H-treated HCMV DNA. When HCMV DNA was treated with RNase H, a 3.2-kb band was detected in addition to the expected 5.9-kb band (Fig. 7A, lane 2). In samples that were not treated with RNase H or in samples where a cosmid containing the HCMV oriLyt region(pCM1029) was treated with RNase H, no additional band was detected. Detection of the 3.2-kb band is consistent with the cleavage of oriLyt DNA near the NotI site at nt 92888 (Fig. 3). However, since RNase H degrades the entire RNA strand, it is not possible to determine the exact cleavage area. RNase H sensitivity was also observed in oriLyt when plasmid pOri was used in the transient-replication assay. pOri was cotransfected along with plasmids containing all of the required replication genes (22). Hirt extracts of plasmid DNA were treated with RNase H. In samples where all of the required genes were present in the cotransfection, RNase H cleavage of pOri was detected (Fig. 7B, lane 1). However, when a plasmid encoding UL105, a protein required for transient replication, was omitted from the transfection mixture, no RNase H-cleaved DNA could be detected (lane 2). Duplicate samples were cleaved with DpnI to ensure that replicated plas-

FIG. 7. vRNA-2 is integrated in the genome as an RNA-DNA hybrid. (A) Viral DNA was treated with RNase H, cleaved with BglII and PvuII as described in Materials and Methods, and separated under denaturing conditions. The samples were hybridized with the subclone pOri SB-B. Lanes: 1, AD169 DNA cleaved with BglII and PvuII; 2, AD169 DNA cleaved with BglII and PvuII and treated with RNase H; 3, cosmid pHC1029 cleaved with BglII and PvuII and treated with RNase H. The arrows indicate the full-length 5.9-kb fragment and the 3.2-kb fragment resulting from an RNase H cleavage. (B) OriLyt plasmid reacted with RNase H after the cotransfection-replication assay. Total-cell DNA was cleaved with HindIII, which cuts pOri only once. Lanes: 1, total-cell DNA from a cotransfection using the required replication genes along with pOri and probed with pOri SB-B; 2, total-cell DNA from a cotransfection using the required replication genes along with pOri, except that the plasmid encoding UL105 was omitted. The arrow indicates a 3.2-kb cleavage product originating near the NotI (nt 92888) site.

7002

PRICHARD ET AL.

J. VIROL.

FIG. 8. HCMV oriLyt region from nt 89796 to 9486. Shown are the relative positions of vRNA-1 and an expanded view of the region where vRNA-2 maps. Also shown is the relative position of the SRT. The shaded areas are regions which were previously shown to be indispensable for origin function. Also shown is the region where an extra 300 bp of DNA sequence is located which is composed of reiterated repeated sequence elements.

mid was present (26). We have performed experiments where we have omitted each replication gene from the transfection mixture, and we did not detect RNase H cleavage in any of these experiments (26). These results indicated that the incorporation of vRNA-2 within oriLyt takes place at the time of or after the replication process in the transient-replication assay but does not demonstrate an absolute requirement for DNA replication in the formation of hybrid structures. Figure 8 is a schematic of HCMV oriLyt indicating the locations of vRNA-1 and vRNA-2. vRNA-2 localizes to a region of oriLyt which exhibits a heterogeneous restriction pattern. As mentioned above, it became apparent while mapping vRNAs that the origin subclone, pOri, contained additional sequence which was not represented in the original HCMV published sequence (5). Mapping experiments had confirmed that this additional sequence was present between the SphI (nt 92576) and BamHI (nt 93513) restriction sites. To define the structure of this region in the context of the viral genome, we probed an SphI-BamHI restriction digest of AD169 DNA with plasmid pOri SB-B. The predicted size of the SphI-BamHI fragment is 650 bp. However, when a SphI-BamHI digest of AD169 DNA was hybridized with pOri SB-B, a series of bands ranging from 850 to 1,000 bp were detected (Fig. 9A). This result indicated that this region of oriLyt was heterogenous with respect to the presence of the restriction sites SphI and BamHI. The heterogeneous section was subcloned from viral DNA by PCR with primers which flank this area. Several PCR products were sequenced, including one which included the entire 1,000-bp region. In addition, fragments cloned directly from viral DNA were sequenced to ensure the accuracy of the sequence derived from the PCR product. Inspection of the DNA sequence from this region of oriLyt revealed that three repeated sequence ele-

ments are reiterated a variable number of times within this region (Fig. 9B). DNA sequencing of the subclone, pOri also showed that it contained the same reiterated sequence elements which corresponded to the extra 300 bp of DNA sequence reported here. DISCUSSION While searching for unusual DNA structures within the HCMV origin of replication, oriLyt, we observed two alkalisensitive loci which flanked an alkali-resistant DNA segment. Initially, we suspected the presence of NaOH-sensitive abasic regions within oriLyt. However, through the development of a procedure to isolate RNA associated with viral DNA, we discovered that alkali-sensitive areas within oriLyt were due to the presence of RNA-DNA hybrid structures. These vRNAs were detected by Northern analysis and were degraded by RNase A treatment. Southern blots of alkali-treated DNA and RNase H experiments revealed that vRNAs were present as stable and persistent RNA-DNA hybrid structures flanking a DNA-DNA segment of oriLyt and are covalently linked to DNA segments. vRNAs are integrated and packaged into virions, as demonstrated by both alkali sensitivity and RNase H cleavage of virion DNA. To ensure that vRNAs were not the result of cellular RNA contamination, purified virus isolated from a sucrose gradient was treated with RNase A. RNase H cleavage experiments indicated that not all packaged viral genomes contained vRNA within oriLyt, in that a percentage of genomic DNA apparently was not cleaved. It is clear, however, that a large percentage of packaged viral DNA was cleaved by RNase H. It is difficult to quantitate the relative amounts of cleavage

VOL. 72, 1998

FIG. 9. The HCMV oriLyt region contains reiterated repeated sequence elements. (A) SphI-BamHI heterogeneous restriction pattern. Shown is an autoradiogram of a Southern blot of AD169 viral DNA cleaved with SphI-BamHI and hybridized with pOri SB-B. (B) Three sequence motifs that are variable reiterated within the SphI and BamHI restriction sites.

since double-stranded probes were used and the uncleaved DNA strand was still detectable in Southern blots. HCMV genomes may contain variable lengths and amounts of vRNAs; in addition, some genomes may contain vRNAs that have secondary structures that make them poor substrates for bacterial RNase H. There is also the possibility that a percentage of viral genomes do not have RNA-DNA hybrids. These viral species may be defective or may not replicate with the same efficiency or by the same mechanism as hybrid-containing species. Various attempts to map the exact 39 end of these vRNAs revealed the absence of a distinct termination site (26). This suggests that vRNAs are integrated with staggered ends during the replication process. Five prime end mapping has been more successful. 59-RACE (rapid amplification of cDNA ends) data indicates that the 59 end of vRNA-2 is about 300 bases upstream of the BamHI site at nt 93513 (26). This is consistent with alkali sensitivity data, where a 500-base region separates the two vRNAs. Southern analysis also shows that vRNA-2 is contained entirely within the SphI-BamHI fragment, a region required for oriLyt function (1). Attempts in our laboratory to retain oriLyt function in constructs where portions of the region between the NotI (nt 92888) and BamHI (nt 93515) site were deleted have failed (26). This suggests that the region where this RNA-DNA hybrid exists is necessary for replication in the transient-replication assay. In addition, although vRNA-1 maps to a region which is dispensable, oriLyt function is severely defective when a deletion to the SacI (nt 93715) site was tested in transient-replication assays (1, 26). vRNA-2 is within a region of oriLyt containing variably

RNA-DNA HYBRID AT oriLyt

7003

repeated sequence elements. An SphI-BamHI restriction digest of AD169 DNA revealed a heterogeneous pattern when hybridized with plasmid pOri SB-B. When subclones of viral DNA were sequenced, the numbers of three repeated sequence elements were variable. These sequence elements accounted for the extra 300 bp observed within the pOri subclone. These elements were not found anywhere else in the AD169 genome and did not have significant homology to any sequence in the GenBank database. The role of these reiterated repeat sequence elements in vRNA formation or in the replication of oriLyt is unknown. In addition, the presence of these variably repeated regions may not fully explain the appearance of a DNA ladder, as seen in Fig. 9, when AD169 DNA was cleaved with SphI and BamHI. This phenomenon could also be due in part to the inefficient cleavage of this region of HCMV DNA because of the presence of various lengths of the RNA component of a RNA-DNA hybrid. This structure would be a poor substrate for restriction endonucleases. Similar types of repeat structures have also been found within the origin of replication of human herpesvirus 6 (HHV-6) (30). In HHV-6, intragenomically amplified sequences within the origin may be the result of the acquisition of oriLyt by transposition (30). The role of these amplified sequences in HHV-6 is not known. Also unknown is if the oriLyt region of HHV-6 contains RNA-DNA hybrid structures. In HCMV, regions containing repeated sequences are necessary for oriLyt function in the transient-replication assay. The vRNA-2 hybrid structure is in close proximity to a recently identified transcript within oriLyt, SRT. In addition, although strand-specific probes were not used to define vRNA-1, this vRNA originates from the same strand as vRNA-2, since alkali treatment was capable of releasing the internal DNA segment. This could occur only if the vRNAs were on the same strand. vRNA-2 is just upstream of SRT and originates from the same DNA strand as SRT. SRT itself was not isolated during the vRNA extraction procedure, indicating that SRT is not embedded within the viral genome. However, vRNA-2 and SRT may be related, in that SRT may be a portion of vRNA-2 not embedded within the HCMV genome. SRT and vRNA-2 could be the result of transcription from an upstream promoter. Multiple transcription factor binding sites exist within oriLyt, suggesting the presence of putative promoter regions. Analysis of regions upstream of vRNA-2 indicate that these regions can act as promoters in transient-replication assays (26). This observation is consistent with models where transcription and/or transcriptional elements play a pivotal role in initiation of DNA synthesis (6, 16, 20). The formation of persistent RNA-DNA hybrid structures as a result of transcription has been described previously (19, 34). In addition, the RNA-DNA hybrid formation at a bacteriophage T4 replication origin has been described (4). In this case, the formation of a persistent RNA-DNA hybrid may be the result of a combination of DNA unwinding and transcription controlled from a middle-mode promoter during replication (4). In T4, the unwinding potential of the downstream promoter region may facilitate RNA-DNA hybrid structure formation by increasing the opportunity for the 59-end of the origin transcript to reassociate with its template strand during transcription (4). The formation of an RNA-DNA hybrid structure was detected in the pOri construct replicated in the transient-replication assay. RNase H cleavage data from this assay suggests that insertion of vRNA-2 is dependent on the inclusion of the DNA replication proteins in the transfection mixture. We cannot determine if RNA-DNA hybrid formation is related to the

7004

PRICHARD ET AL.

activity of UL105 per se or to the transient replication of the plasmid itself. Nevertheless, the presence of a RNA-DNA hybrid structure suggests that HCMV DNA replication may involve an RNase H-like mechanism. If this is the case, an enzyme with this function must be a component of one of the required replication genes. One such virus-encoded candidate is the UL84 gene product (13). Initially, UL84 was shown to be 1 of 11 loci required for oriLyt-dependent DNA replication (22). More recently, this observation was underscored by Sarisky and Hayward, who demonstrated that UL84 plays a critical role in transient replication (27). The exact function of the UL84 protein in DNA replication remains unknown. However, in transient-replication assays, we have observed that the UL84 gene product has an effect on RNA stability (26). How might vRNAs play a role in HCMV DNA replication? vRNAs may function as distinct areas where initiation of DNA replication occurs. The first step might be RNA-DNA hybrid formation resulting from transcription originating from an upstream promoter. RNA-DNA hybrid structures could act as substrates for a specific virus-encoded RNase H-type enzyme. This virus-encoded enzyme would cause a nick or partial degradation of the RNA strand of the RNA-DNA hybrid. The remaining RNA portion, or the free DNA end resulting from the removal of integrated RNA, would function as a primer from which DNA replication would initiate. Consistent with this mode of replication, the RNA strand would be incorporated into a percentage of viral genomes as a consequence of the normal replication process. This in turn would ensure the initial round of replication upon new infection independent of a new transcription event. This type of replication scheme suggests that HCMV DNA replication takes place on two levels. The first would occur in viral templates where RNADNA hybrids already exist, and the second would be dependent on a transcriptional event that would cause the formation of the hybrid structure. The presence of an RNA-DNA hybrid in oriLyt may not be an absolute requirement for DNA replication; however, they may facilitate an early or first phase of DNA synthesis. An HCMV delayed-DNA synthesis phenotype has been described, suggesting that HCMV DNA replication may occur in two stages (24, 25). We are currently investigating the function and mechanism of insertion of vRNAs together with exploring the role of UL84 and its association, if any, with vRNAs. REFERENCES 1. Anders, D. G., M. A. Kacica, G. S. Pari, and S. M. Punturieri. 1992. Boundaries and structure of human cytomegalovirus oriLyt, a complex origin for lytic-phase DNA replication. J. Virol. 66:3373–3384. 2. Anders, D. G., and S. M. Punturieri. 1991. Multicomponent origin of cytomegalovirus lytic-phase DNA replication. J. Virol. 65:931–937. 3. Bruckner, R. C., J. J. Crute, M. S. Dodson, and I. R. Lehman. 1991. The herpes simplex virus 1 origin binding protein: a DNA helicase. J. Biol. Chem. 266:2669–2674. 4. Charles-Kinch, K., and K. N. Kreuzer. 1997. RNA-DNA hybrid formation at a bacteriophage T4 replication origin. J. Mol. Biol. 266:915–926. 5. Chee, M. S., A. T. Bankier, S. Beck, R. Bohni, C. M. Brown, R. Cerny, and T. Horsnell. 1990. Analysis of the coding content of the sequence of human cytomegalovirus strain AD169. Curr. Top. Microbiol. Immunol. 154:125– 169. 6. DePamphilis, M. L. 1988. Transcriptional elements as components of eukaryotic origins of DNA replication. Cell 52:635–638. 7. Dewhurst, S., S. C. Dollard, P. E. Pellett, and T. R. Dambaugh. 1993. Identification of a lytic-phase origin of DNA replication in human herpesvirus 6B strain Z29. J. Virol. 67:7680–7683. 8. Dewhurst, S., D. M. Krenitsky, and C. Dykes. 1994. Human herpesvirus 6B origin: sequence diversity, requirement for two binding sites for originbinding protein, and enhanced replication from origin multimers. J. Virol. 68:6799–6803. 9. Elias, P., C. M. Gustafsson, and O. Hammarsten. 1990. The origin binding protein of herpes simplex virus 1 binds cooperatively to the viral origin of

J. VIROL. replication Oris. J. Biol. Chem. 265:17167–17173. 10. Fixman, E. D., G. S. Hayward, and S. D. Hayward. 1995. Replication of Epstein-Barr virus oriLyt: lack of a deticated virally encoded origin-binding protein and dependence on Zta in cotransfection assays. J. Virol. 69:2998– 3006. 11. Hammerschmidt, W., and B. Sugden. 1988. Identification and characterization of oriLyt, a lytic origin of DNA replication of Epstein-Barr virus. Cell 55:427–433. 12. Hamzeh, F. M., P. S. Lietman, W. Gibson, and G. S. Hayward. 1990. Identification of the lytic origin of DNA replication in human cytomegalovirus by a novel approach utilizing ganciclovir-induced chain termination. J. Virol. 64:6184–9195. 13. He, Y. S., L. Xu, and E.-S. Huang. 1992. Characterization of human cytomegalovirus UL84 early gene and identification of its putative protein product. J. Virol. 66:1098–1108. 14. Huang, L. L., Y. Zhu, and D. G. Anders. 1996. The variable 39 ends of a human cytomegalovirus oriLyt transcript (SRT) overlap an essential, conserved replicator element. J. Virol. 70:5272–5281. 15. Kroff, A., J. F. Schwedes, and P. Tegtmeyer. 1991. Herpes simplex virus origin-binding protein (UL9) loops and distorts the viral replication origin. J. Virol. 65:3284–3292. 16. Li, R., and M. R. Botchan. 1994. Acidic transcription factors alleviate nucleosome-mediated repression of DNA replication of bovine papillomavirus type 1. Proc. Natl. Acad. Sci. USA 91:7051–7055. 17. Lucin, P., S. Jonjic, M. Messerle, B. Polic, H. Hengel, and U. H. Koszinowski. 1994. Late phase inhibition of murine cytomegalovirus replication by synergistic action of interferon-gamma and tumour necrosis factor. J. Gen. Virol. 75:101–110. 18. Masse, M. J. O., S. Karlin, G. A. Schachtel, and E. S. Mocarski. 1992. Human cytomegalovirus origin of replication (oriLyt) resides within a highly complex repetitive region. Proc. Natl. Acad. Sci. USA 89:5246–5250. 19. Masukata, H., and J. Tomizawa. 1990. A mechanism of formation of a persistent hybrid between elongating RNA and template DNA. Cell 62:331– 338. 20. Mohr, I. J., R. Clark, S. Sun, E. J. Androphy, P. MacPherson, and M. R. Botchan. 1991. Targeting the E1 replication protein to the papillomavirus origin of replication by complex formation with the E2 transactivator. Science 250:1694–1699. 21. Olivo, P. D., N. J. Nelson, and M. D. Challberg. 1988. Herpes simplex virus DNA replication: the UL9 gene encodes an origin-binding protein. Proc. Natl. Acad. Sci. USA 85:5414–5418. 22. Pari, G. S., and D. G. Anders. 1993. Eleven loci encoding trans-acting factors are required for transient complementation of human cytomegalovirus oriLyt-dependent DNA replication. J. Virol. 67:6979–6988. 23. Pari, G. S., M. A. Kacica, and D. A. Anders. 1993. Open reading frames UL44, IRS1/TRS1, and UL36-38 are required for transient complementation of human cytomegalovirus oriLyt-dependent DNA synthesis. J. Virol. 67:2575–2582. 24. Prichard, M. N., G. M. Duke, and E. S. Mocarski. 1996. Human cytomegalovirus uracil DNA glycosylase is required for the normal temporal regulation of both DNA synthesis and viral replication. J. Virol. 70:3018–3025. 25. Prichard, M. N., N. Gao, E. S. Mocarski, D. Yu, S. Jairath, and G. S. Pari. Human cytomegalovirus uracil DNA glycosylase participates in the originspecific initiation of viral DNA synthesis and associated with ppUL44. Submitted for publication. 26. Prichard, M. N., and G. S. Pari. Unpublished data. 27. Sarisky, R. T., and G. S. Hayward. 1996. Evidence that the UL84 gene product of human cytomegalovirus is essential for promoting oriLyt-dependent DNA replication and formation of replication compartments in cotransfection assays. J. Virol. 70:7398–7413. 28. Schepers, A. D., D. Pich, J. Mankertz, and W. Hammerschmidt. 1993. cisacting elements in the lytic origin of DNA replication of Epstein-Barr virus. J. Virol. 67:4237–4245. 29. Smith, J. A., and G. S. Pari. 1995. Human cytomegalovirus UL102 gene. J. Virol. 69:1734–1740. 30. Stamey, F. R., G. Dominguez, J. B. Black, T. R. Dambaugh, and P. E. Pellett. 1995. Intragenomic linear amplification of human herpesvirus 6B oriLyt suggests acquisition of oriLyt by transposition. J. Virol. 69:589–596. 31. Stow, N. D. 1982. Localization of an origin of DNA replication within the TRs/IRs repeated region of the herpes simplex virus type 1 genome. EMBO J. 1:863–867. 32. Stow, N. D., and A. J. Davidson. 1986. Identification of varicella-zoster virus origin of DNA replication and its activation by herpes simplex virus type 1 gene products. J. Gen. Virol. 67:1613–1623. 33. Weir, H., and N. D. Stow. 1990. Two binding sites for the herpes simplex type 1 UL9 protein are required for efficient activity of the oris replication origin. J. Gen. Virol. 71:1379–1385. 34. Xu, B., and D. A. Clayton. 1995. A persistent RNA-DNA hybrid is formed during transcription at a phylogenetically conserved mitochondrial DNA sequence. Mol. Cell. Biol. 15:580–589.