Human Cytomegalovirus with IE-2 (UL122) Deleted Fails To Express Early Lytic Genes

Cells, viruses, viral DNA, and cosmid clones.

Primary human foreskin fibroblast (HFF) cells were prepared from tissue samples and grown in Dulbecco's modified Eagle medium plus 10% fetal calf serum. The viral DNA used to create a BAC was purified from total virus particles isolated from HFF cells infected with the Towne strain of HCMV (a gift from T. Shenk) according to established protocols (19). Cosmid subclones comprising the entire AD169 HCMV genome (a gift from B. Fleckenstein and T. Shenk) were used to confirm the structure of the Towne BAC (T-BAC) by Southern blotting.

Production and characterization of a HCMV BAC.

To provide flanking DNA for homologous recombination in eukaryotic cells, two fragments of HCMV DNA were PCR amplified from cosmid clone CM1052, which contains the HindIII K/Q, X, V, and W fragments of AD169 HCMV (12). Amplifications used the following primers (read 5′ to 3′) derived from the published sequence (9) of AD169 HCMV DNA:(i) CCGGATCCCCACCGGGTAGAACC, in which the first cytosine residue following the BamHI site is nucleotide 189941; and (ii) CCAAGCTTGCACAACGGGATGACC, in which the guanosine residue following the HindIII site is the complement of nucleotide 191921. Primers i and ii yielded a 1.98-kb PCR product and introduced restriction sites for BamHI and HindIII at the 5′ and 3′ termini, respectively. A second recombination fragment was amplified similarly, using the following primers: (iii) GAGCCAGAGTATGGG, in which the first residue is nucleotide 200834 and occurs just 5′ to the HindIII site starting at nucleotide 200856; and (iv) CCTATCTACGTGCCC, in which the last cytosine is the complement of nucleotide 204034 and occurs just 3′ to the BamHI site starting at nucleotide 204024. Primers iii and iv yielded a 3.2-kb product including unique BamHI and HindIII sites near the termini.

The F-plasmid vector pMBO1374, a gift from G. Smith (34), was first modified by removal of the unique BamHI site and one of two ClaI sites, giving pMBO1374.1. The two HCMV PCR products (see above) were digested with BamHI and ligated to each other, resulting in a 5.2-kb linear species with the 3′ end of the 3′ homology fragment joined to the 5′ end of the 5′ fragment (Fig. 1A). The 5.2-kb fragment was next digested with HindIII and cloned into the HindIII site of pMBO1374.1, yielding plasmid pUSF-2. A cassette in which the simian virus 40 early promoter and polyadenylation signals control expression of green fluorescent protein (GFP) was PCR amplified from plasmid pGET-07 (a gift from G. Tullis) and cloned into the remaining ClaI site of pUSF-2 via ClaI sites included in the primers. This final construct, pUSF-3, contains the prokaryotic genetic elements necessary to confer maintenance as a BAC in E. coli, HCMV DNA sequences to direct homologous recombination to the unique short (US) region of the viral genome, and the GFP marker to facilitate identification and purification of recombinant HCMV in eukaryotic cells. The flanking DNA deletes 8.9 kb of DNA within the US region of HCMV that has been defined as dispensable for HCMV replication in cell culture (18), truncating IRS1 after amino acid 719 and removing reading frames US1 to US11 plus the carboxy-terminal third of US12.

To generate the recombinant HCMV incorporating the BAC and GFP sequences, pUSF-3 was digested with BamHI and electroporated into HFF cells along with wild-type Towne viral DNA and an expression plasmid for the HCMV tegument protein pp71 (2). One day following transfection, the surviving cells were distributed into 96-well culture dishes, and these were monitored for development of cytopathic effect. When cytopathic effect was uniform in the cultures (10 to 14 days), the supernatants were collected from 12 independent wells in which the majority of cells expressed GFP, and virus carrying the GFP marker was enriched by plaque purification. To recover the recombinant HCMV as covalently closed circular BAC DNA, HFF cells were infected at high multiplicity of infection (MOI) with the enriched stocks for 24 h, after which a crude lysate was prepared from the infected monolayer by lysis at 55°C in Tris-HC1 (50 mM, pH 8.0), EDTA (50 mM), sodium dodecyl sulfate (1%), and proteinase K (100 μg/ml). Aliquots of this crude lysate were transformed, without further purification, into RecA⁻ E. coli DH10B (Research Genetics, Inc.), and transformants resistant to chloramphenicol were selected. As an initial screen for a CMV BAC having the expected structure, Towne viral DNA and BAC DNA from resistant colonies were digested with EcoRI and electrophoresed in agarose, and the fragments were visualized with ethidium bromide. Out of approximately 100 transformants examined, one displayed a restriction pattern nearly identical to that of Towne viral DNA. This BAC clone, designated T-BACwt, was further analyzed by Southern blotting with cosmid probes that spanned the entire AD169 genome.

To characterize the virus which was reconstituted following T-BACwt transfection, cells and supernatant were collected from HFF cultures transfected with either T-BACwt DNA or Towne DNA purified from virions, subjected to two cycles of freeze-thawing, cleared, and titered on fresh HFF cells. To compare the growth kinetics of Towne and T-BACwt viruses, HFF cells were infected with 0.01 PFU/cell at 37°C for 1 h, residual virus was washed out, and the cells were trypsinized and divided among six dishes. Dishes were harvested at various times postinfection, and the progeny virus was titered on fresh HFF cells as described above.

Transfer constructs and conjugative transfer.

The procedure and reagents for conjugative transfer of sequences to BAC DNA in E. coli were generously supplied by G. Smith and L. Enquist and have been described elsewhere (11, 34). To create a transfer vector to mutate UL122 from HCMV, a deletion of approximately 1,230 bp was introduced into an EcoRI-SalI restriction fragment subclone of the Towne MIE region, from the SmaI site at the 5′ end of the UL122 reading frame to the StuI site at the 3′ end (relative to transcription of the gene). This allele was cloned into pGS284, a derivative of the suicide selection vector pCVD442 (34). This allele produces an IE-2 polypeptide with a frameshift after amino acid 135 followed by truncation after an additional 20 unrelated amino acids; it thus contains the 85 residues common to IE-1 and IE-2 plus 50 amino-terminal residues of exon 5 unique to IE-2. A wild-type allele of the IE-2 region was cloned independently into pGS284 and used to rescue the UL122 deletion BAC, also by conjugative transfer. Briefly, to transfer DNA sequences in pGS284 to the HCMV BAC, E. coli S17-λpir containing the GS284 donor plasmid was conjugated with a RecA⁺ derivative of E. coli DH10B (34) harboring the HCMV BAC DNA. Exconjugates were selected sequentially with antibiotics and sucrose, and the progeny molecules were examined by restriction digestion to identify BAC clones with the intended alteration.

Transfection by electroporation.

Actively dividing HFF cells were trypsinized and suspended in Dulbecco's modified Eagle medium with 10% fetal calf serum. For each transfection, 4 × 10⁶ HFF cells were suspended in 260 μl of medium plus 10% serum and mixed with 2 μg of viral or BAC DNA, 1 μg of plasmid pCMV71 (23) (a gift from J. Baldick), and 1 μg of plasmid pEGFP-N1 (Clontech) in a 0.4-cm cuvette. Following electroporation at 250 V and 960 μF, the cells were plated in a 10-cm-diameter tissue culture plate. Typically, 1 to 2 days after transfection the surviving cells were approaching confluence and were split between 1:2 and 1:4 into new dishes. The passaged cells were cultured at 37°C for plaque outgrowth or harvested for analysis. Plaques were scored on the criteria of visual morphology or GFP expression within multiple adjacent cells in the monolayer. The pattern and spread of GFP fluorescence to neighboring cells in this assay are similar to those observed by IE-1 or IE-2 staining of cultures infected with wild-type Towne virus (2), although GFP expression is delayed by several days relative to IE gene expression.

Molecular analyses, RT-PCR, and immunofluorescence.

Plasmid cloning, restriction enzyme digestion, gel electrophoresis, PCR, and Southern blotting were performed according to established protocols (1). For reverse transcription-PCR (RT-PCR) analysis, cells were transfected with BAC DNA and plasmids as described above. To maximize our ability to detect HCMV gene expression from transfected BAC DNA, cultures in which at least 10 to 20% of the cells expressed GFP 4 to 6 days after transfection were used for RT-PCR analysis. Total RNA was isolated using TRIZOL reagent (Gibco BRL) and digested with RNase-free DNase (Promega). First-strand cDNA was synthesized from an oligo(dT) primer using PowerScript reverse transcriptase (Clontech), according to the manufacturer's protocol. This cDNA served as the template for PCR (35 cycles) using Advantage cDNA polymerase (Clontech) and primer sets for various HCMV IE and early genes as indicated in text and figure legends. PCR products were analyzed by agarose gel electrophoresis.

Expression of HCMV IE proteins in transfected cultures was analyzed by immunofluorescence using monoclonal antibodies 1B12 and 3H9, specific for the unique domains of IE-1 and IE-2, respectively (Marchini and Zhu, unpublished). Briefly, cultures were fixed 3 days after transfection in 4% paraformaldehyde, permeabilized, and blocked with bovine serum albumin. The fixed cells were reacted with primary monoclonal hybridoma culture supernatants followed by secondary antibody conjugated to Alexa-568 (Molecular Probes). Stained cultures were examined by inverted epifluorescence microscopy.

Production of a Towne HCMV BAC.

A Towne HCMV BAC was produced and isolated essentially as described by Borst et al. (4), beginning with introduction of BAC sequences into the viral genome via homologous recombination in cultured mammalian cells and then recovery and maintenance of circular genomes in E. coli. To avoid creating a recombinant genome too large to package, we designed a plasmid vector, pUSF-3 (see Fig. 1 and Materials and Methods), that would delete about 9 kb of HCMV DNA from the dispensable portion of the US region (18). Transfer of pUSF-3 to the Towne US target yields a recombinant HCMV containing F-plasmid sequences necessary for episomal maintenance in E. coli, a selectable marker conferring prokaryotic resistance to chloramphenicol, and a cassette for eukaryotic expression of GFP (Fig. 1A).

Following electroporation of RecA⁻ E. coli with closed circular DNA from infected HFF cells, BAC DNA was prepared from resistant colonies and examined by restriction digestion. A clone with a restriction pattern comparable to that of Towne viral DNA was examined extensively by Southern blotting (Fig. 2). Initially, blots were probed with each of the HCMV DNA fragments in pUSF-3 to examine the structure of the US region to which the recombination had been targeted (Fig. 2A). A Southern blot probed with the 3-kb pUSF-3 fragment showed the 15.9-kb EcoRI C fragment of Towne replaced by the predicted 12.5-kb fragment in the recombinant (Fig. 2B, lanes 1 and 2). Probing with the 2-kb pUSF-3 fragment showed the predicted reduction of the 10.5-kb EcoRI U/M junction fragment to 7.5 kb (Fig. 2B, lanes 3 and 4). The 2-kb probe contains DNA from the terminal repeat of the US region and thus detected all fragments containing either end of that region. One of these, the 7-kb EcoRI U/Z junction fragment, occurs in both the BAC and viral DNAs (lanes 3 and 4). Consistent with the anticipated lack of linear viral DNA in the BAC, the 2-kb probe detects the 6.5-kb EcoRI M and 3-kb EcoRI Z free-end fragments only in viral DNA (lane 3). The origin of the additional bands of 3.7 and 4.4 kb visible in our Towne viral DNA preparation (lane 3) is unclear; the sizes are consistent with EcoRI-Z fragments that have one and two additional copies, respectively, of the 700-bp terminal a sequence and may arise from differential cleavage between tandem copies of the terminal sequence during cleavage of concatemeric DNA (24). The pMBO1374 BAC vector was also used as a probe and detected an internal 8.9-kb EcoRI fragment from the integrated plasmid in BAC DNA but failed to hybridize to viral DNA (compare lanes 5 and 6). Additional Southern blot analyses with cosmid probes that span the entire HCMV genome further confirmed that this BAC clone contained, without rearrangement, all of the EcoRI fragments present in the parental Towne viral DNA (data not shown). The restriction mapping analysis also indicated that the genome in the BAC clone was in the prototypical isomeric arrangement, consistent with the structure of pUSF-3. As observed with the AD169 HCMV BAC clone, there was no evidence of genome isomerization within E. coli (4). These analyses confirmed all predictions for the structure of a recombinant Towne HCMV-BAC resulting from homologous recombination with pUSF-3; therefore, we examined this clone, designated T-BACwt, further to characterize its biological properties.

To establish that our T-BAC clone retained the fundamental biological characteristics of Towne viral DNA, T-BACwt DNA purified from E. coli was transfected into HFF cells with an expression plasmid for the HCMV tegument protein pp71 (23). The transfected cells were plated, and the cultures were monitored for development of cytopathic effect and peak virus titer (Fig. 3). Wild-type Towne DNA, purified from virions, and T-BACwt DNA from E. coli both produced plaques following transfection into HFF cells (Fig. 3B). Comparing these two molecules in numerous transfections, we found viral DNA to yield more plaques than the BAC, and vice versa; thus, we feel that the observed difference between viral DNA and T-BACwt is due more to variation inherent to these experiments than to a genuine difference in biological features. The peak titers, of infectious virus particles recovered from transfection supernatants were also similar (Fig. 3B). In addition to characterizing production of virus from transfected BAC DNA, we examined the infectious properties of the resulting HCMV BAC virus stocks following low-MOI infection of HFF cells. Towne and T-BACwt viruses generated very similar yields of infectious progeny (Fig. 3A), a minor difference being that T-BACwt, either as transfected DNA or as infectious T-BAC virus, developed visible cytopathic effect more slowly than its wild-type Towne counterpart (data not shown). We do not know whether this results from the deletion of US genes or from a secondary mutation elsewhere in T-BACwt DNA. Aside from this minor phenotypic difference in cytopathic effect, the additional tests established that the T-BACwt bacterial clone of Towne HCMV retains the essential properties of wild-type Towne virus; therefore, T-BACwt was taken as the starting reagent for subsequent experiments.

Derivation of a deletion mutant of IE-2.

Our primary motivation for creating an HCMV BAC was a long-standing interest in genetic analysis of the MIE gene IE-2 (pp86). Our inability to isolate an IE-2 deletion mutant free of contaminating wild-type virus by plaque purification on normal HFF cells suggested first that this mutant was impaired for growth (Marchini and Zhu unpublished) and additionally that available means for providing complementing IE-2 were inadequate. The IE-1 and IE-2 transcripts share three small 5′ exons and then splice differentially to either of two unrelated exons that encode the majority of each protein (Fig. 4A). To create a mutation which would have maximal effect on IE-2 without directly affecting IE-1, the majority of the main exon unique to IE-2, encoded by the UL122 reading frame, was deleted from a 6.7-kb EcoRI-SalI subclone of the Towne MIE region. The deleted allele with flanking sequences for recombination was cloned into pGS284, a derivative of the positive suicide selection vector pCVD442 (see Materials and Methods), and conjugated into RecA⁺ E. coli harboring T-BACwt. Exconjugates were selected, and the T-BAC DNA was examined by Southern blotting. Approximately half of the exconjugate T-BAC clones now had a 5.5-kb EcoRI-SalI MIE fragment that was detected by a wild-type MIE probe but failed to hybridize to the deleted SmaI-StuI exon 5 fragment, as would be expected from a T-BAC clone containing the UL122 deletion (clones M1 and M2 in Fig. 4B to D). Two of these were selected for further use and were designated T-BACΔ122. To control for mutations at unrelated loci that might have been introduced during production of the T-BACΔ122 genome, the two T-BACΔ122 clones were each rescued to the wild-type state by repeating the conjugative mating protocol with a wild-type UL122 donor allele. Again, about half of the clones examined following selection had the UL122 locus repaired, as evidenced by the return of the EcoRI-SalI MIE fragment to the 6.7-kb wild-type size, and restored hybridization to the SmaI-StuI exon 5 probe (representative clone R2 in Fig. 4B to D). As is typical with conintegrate resolution, those genomes that did not incorporate the mutation retained the structure of the target BAC. An example of this is the rescue clone labeled R1 in Fig. 4. The two original T-BACΔ122 clones and one repaired derivative of each (designated T-BACΔ122R) were used for the experiments described below.

A UL122 deletion BAC is not infectious.

To begin assessing the biological properties of the UL122 mutant, we electroporated purified T-BACwt, T-BACΔ122, and T-BACΔ122R DNAs into permissive HFF cells and monitored the transfected cultures for plaque development. The transfection experiments were repeated many times with multiple viral DNA and BAC DNA preparations to minimize the chance that negative results were due to defects in the BAC DNA introduced during passage in E. coli or DNA preparation. After 7 to 10 days, nascent plaques were identified by GFP expression, and the total plaque yield from each transfected DNA was scored. Selected representative results are presented in Table 1. Experimental variation, due primarily to differences in transfection efficiency and the quality of individual BAC DNA preparations, produced a wide range of plaque yields for T-BACwt and T-BACΔ122R, from a few to several hundred per transfection; thus, a critical quantitative interpretation of these data is of limited value. Nonetheless, qualitative examination shows the clear difference between these substrates and T-BACΔ122 DNA, which never yielded a recognizable plaque either in the six independent transfections reported in Table 1 or in 15 additional experiments (not shown).

To confirm that the mutation gave the expected phenotype at the protein level, cultures transfected with BAC DNA were fixed and stained using monoclonal antibodies specific for IE-1 or IE-2 expression. With GFP expression from a cotransfected plasmid as a guide to identify transfected cells, expression of IE-1 was readily detected from either T-BACΔ122 or T-BACΔ122R genomes 4 to 5 days after transfection (Fig. 5, top row). In contrast, expression of IE-2 was observed only in cultures transfected with T-BACΔ122R DNA; no IE-2 protein was found in any cells transfected with T-BACΔ122 DNA (Fig. 5, bottom row). The monoclonal antibody used to detect IE-2 was specific to an epitope within the deletion (A. Marchini, unpublished observations); thus, we do not know whether the truncation protein predicted to be synthesized from the deletion allele (see Materials and Methods) was produced in T-BACΔ122-transfected cells.

The failure to recover a plaque from 21 independent transfection experiments strongly suggests that the inability of T-BACΔ122 to produce plaques is not a chance occurrence but rather is due to the deficiency of IE-2. Even when T-BACΔ122-transfected HFF cells were cultured for 4 weeks after transfection, we found no evidence of plaque formation as manifested by either visible cytopathic effect or GFP expression. The possibility that the replication defect is due to mutation at a secondary site within the BAC is effectively ruled out by our ability to rescue the phenotype of the mutant with wild-type UL122 sequences. The all-or-nothing nature of this assay, unfortunately, prevents our making any substantial quantitative statement regarding the degree of impairment relative to the control viruses, and since we have been unable to complement the UL122 deletion mutant BAC, we cannot yet generate stocks of mutant virus with which to perform quantitative virological assessments.

The UL122 deletion mutant is defective for early gene expression.

Detection of IE-1 expression from transfected T-BACΔ122 DNA by immunofluorescence indicated that aspects of infection prior to IE gene expression were not markedly impaired in the UL122 mutant. We therefore examined the expression of other IE loci and a panel of early genes using RT-PCR to establish whether these were expressed normally in the mutant. HFF cells were transfected with T-BACΔ122 or T-BACΔ122R DNA and cultured for 5 days, after which total RNA was prepared from the cultures and subjected to RT-PCR analysis using primers to three IE genes and five early genes. RNA expression was detected for IE-1, IE-2, and TRS1 in cells transfected with wild-type Towne (data not shown) and T-BACΔ122R DNA and for IE-1 and TRS1 in the T-BACΔ122-transfected cultures (Fig. 6A). The amplification conditions for this analysis were not quantitative, and therefore the results do not rule out the possibility that IE gene expression is altered in the mutant; however, the results clearly indicate that expression of IE genes, other than IE-2, remained detectable from the T-BACΔ122 genome. In contrast, no expression was detected from the mutant genome for any of the five early genes examined (Fig. 6B). Notably, our assay failed to detect mRNA from TRL4, one of the most abundant early transcripts of HCMV (38). These results, together with the plaque outgrowth experiments in Table 1 demonstrate that HCMV DNA lacking UL122 is growth impaired in HFF cells and that this impediment correlates with a failure to activate expression of early genes.