Complex DNA viruses, with genomes that may express more than 200 gene products, encode collections of proteins that cooperate to perform or significantly modify processes normally executed by the host cell machinery, such as DNA synthesis and regulation of transcription, as well as other processes unique to viral amplification, such as capsid assembly and DNA packaging. These complex virus systems pose major challenges for analysis by classical genetic methods, because of the large number of mutants that must be produced and characterized to identify all of the components involved in such multifactor processes. This chapter describes the general strategy of some approaches, details the methods by which authors have applied them to the problems of defining essential components of the human cytomegalovirus DNA replication and transcription regulating machineries, and, finally, highlights some other applications and related approaches that illustrate the general applicability of transient genetics.

ABSTRACT Cyclooxygenase-2 (COX-2) is a cellular enzyme in the eicosanoid synthetic pathway that mediates the synthesis of prostaglandins from arachidonic acid. The eicosanoids function as critical regulators of a number of cellular processes, including the acute and chronic inflammatory response, hemostasis, and the innate immune response. Human cytomegalovirus (HCMV), which does not encode a viral COX-2 isoform, has been shown to induce cellular COX-2 expression. Importantly, although the precise role of COX-2 in CMV replication is unknown, COX-2 induction was shown to be critical for normal HCMV replication. In an earlier study, we identified an open reading frame (Rh10) within the rhesus cytomegalovirus (RhCMV) genome that encoded a putative protein (designated vCOX-2) with high homology to cellular COX-2. In the current study, we show that vCOX-2 is expressed with early-gene kinetics during RhCMV infection, resulting in production of a 70-kDa protein. Consistent with the expression of a viral COX-2 isoform, cellular COX-2 expression was not induced during RhCMV infection. Finally, analysis of growth of recombinant RhCMV with vCOX-2 deleted identified vCOX-2 as a critical determinant for replication in endothelial cells.

The human cytomegalovirus (HCMV) UL57 gene lies adjacent to HCMV oriLyt, from which it is separated by an organizationally conserved, mostly noncoding region that is thought to both regulate UL57 expression and activate oriLyt function. However, the UL57 promoter has not been studied. We determined the 5' ends of UL57 transcripts toward an understanding of the potential relationship between UL57 expression and oriLyt activation. The results presented here identified three distinct 5' ends spread over 800 bp, at nt 90302, 90530, and 91138; use of these sites exhibited differential sensitivity to phosphonoformic acid treatment. Interestingly, a 10-kb UL57 transcript accumulated in cycloheximide-treated infected cells, even though other early transcripts were not detectable. However, the 10-kb transcript did not accumulate in cells treated with the more stringent translation inhibitor anisomycin. Consistent with the notion that the identified 5' ends arise from distinct transcription start sites, the sequences upstream of sites I and II functioned as promoters responsive to HCMV infection in transient assays. However, the origin-proximal promoter region III required downstream sequences for transcriptional activity. Mutation of candidate core promoter elements suggested that promoter III is regulated by an initiator region (Inr) and a downstream promoter element. Finally, a 42-bp sequence containing the candidate Inr activated a minimal oriLyt core construct in transient replication assays. Thus, these studies showed that a large, complex promoter region with novel features controls UL57 expression, and identified a sequence that regulates both UL57 transcription and oriLyt activation.

The complete DNA sequence of rhesus cytomegalovirus (RhCMV) strain 68-1 was determined with the whole-genome shotgun approach on virion DNA. The RhCMV genome is 221,459 bp in length and possesses a 49% G+C base composition. The genome contains 230 potential open reading frames (ORFs) of 100 or more codons that are arranged colinearly with counterparts of previously sequenced betaherpesviruses such as human cytomegalovirus (HCMV). Of the 230 RhCMV ORFs, 138 (60%) are homologous to known HCMV proteins. The conserved ORFs include the structural, replicative, and transcriptional regulatory proteins, immune evasion elements, G protein-coupled receptors, and immunoglobulin homologues. Interestingly, the RhCMV genome also contains sequences with homology to cyclooxygenase-2, an enzyme associated with inflammatory processes. Closer examination identified a series of candidate exons with the capacity to encode a full-length cyclooxygenase-2 protein. Counterparts of cyclooxygenase-2 have not been found in other sequenced herpesviruses. The availability of the complete RhCMV sequence along with the ability to grow RhCMV in vitro will facilitate the construction of recombinant viral strains for identifying viral determinants of CMV pathogenicity in the experimentally infected rhesus macaque and to the development of CMV as a vaccine vector.

The human cytomegalovirus (HCMV) UL70, UL102, and UL105 genes are predicted to encode essential proteins that assemble the replicative helicase–primase complex based on sequence and genome position similarities to putative herpes simplex virus type 1 (HSV-1) counterparts. Consistent with this prediction, they are required for transient complementation of DNA synthesis. However, little is known about their physical interactions and biochemical activities, primarily because of their restricted expression in HCMV-infected cells. To look for assembly of the predicted complexes, we prepared rabbit polyclonal antisera and used Semliki Forest Virus (SFV) vectors to express untagged and glutathione-S-transferase (GST)-tagged UL70, UL102 and UL105 proteins. The UL70 and UL105 proteins co-purified with the GST-tagged UL102 protein from triply-infected baby hamster kidney cells (BHK-21), and pUL70, but not pUL105, co-purified with pGST–UL102 from dually infected BHK-21 cells. In immunoprecipitation experiments with untagged SFV-expressed proteins, pUL70 or pUL105 coprecipitated with pUL102, pUL102 or pUL70 co-precipitated with pUL105; and pUL102 or pUL105 coprecipitated with pUL70. Comparison of the GST-pull down and immunoprecipitation experiments suggested that the amino-terminal GST-tag interfered with certain pairwise interactions. These results support the prediction that the HCMV helicase–primase proteins assemble a three-protein heteromeric complex, and show that each protein contacts both partners.

Several of the viral proteins required for human cytomegalovirus (HCMV) DNA replication have been difficult to study due to their low abundance in infected cells and low solubility in bacterial or insect-cell expression systems. Therefore we used the Semliki Forest virus expression system to express these proteins in mammalian cells. All of the recombinant proteins were soluble, on the basis of ultracentrifugation properties and their ability to be immunoprecipitated from solution with specific antibodies. Pulse-chase analysis of the 86-kDa major immediate-early protein (IE86) revealed two expressed forms-a precursor and a product-indicating that this recombinant protein, like the native HCMV protein, is posttranslationally processed. The recombinant proteins (polymerase core and accessory as well as the IE86 and pUL84) formed stable complexes similar to those known to form in HCMV-infected cells. The recombinant DNA polymerase holoenzyme also exhibited enzyme activity that was phosphonoformic acid sensitive, as is the infected-cell DNA polymerase activity. This expression system offers many advantages for the expression and study of the HCMV replication proteins, including the expression of soluble, active proteins that are able to interact to form complexes. Additionally, the relative ease with which SFV recombinants can be made lends itself to the construction and evaluation of mutants.

The mechanisms of action and regulation of the human cytomegalovirus (HCMV) lytic-phase DNA replicator, oriLyt, which spans more than 2 kbp in a structurally complex region near the middle of the unique long region (UL), are not understood. Because oriLyt is thought to be essential for promoting initiation of lytic DNA synthesis and may participate in regulating the switch between lytic and latent phases, we undertook a mutational study to better define its sequence requirements. Kanr gene cassette insertions located an oriLyt core region between nucleotides (nt) 91751 and 93299 that is necessary but not sufficient for replicator activity in transient assays. In contrast, insertions into auxiliary regions flanking either side of this core-also required for significant replicator activity-had little effect. To search for essential components within the core region, we made a series of overlapping, roughly 200-bp deletions, and qualitatively and quantitatively assessed the abilities of the resulting constructs to mediate replication. All but one of these deletions produced a significant (i.e., greater than twofold) loss of activity, arguing that sequences across this entire region contribute to replicator function. However, two particularly critical segments separated by a dispensable region, here called essential regions I and II, were identified. Within essential region I, which overlaps the previously identified early transcript SRT, two adjacent but nonoverlapping, roughly 200-bp deletions abolished detectable replication. No single element or motif from the left half of essential region I was found to be essential. Thus, essential region I probably promotes replication through the cooperation of multiple elements. However, four small deletions in the right half of essential region I, which included or lay adjacent to the conserved 31-nt oligopyrimidine tract (referred to as the Y block), abolished or virtually abolished oriLyt activity. Together, these results identify candidate oriLyt sequences within which molecular interactions essential for initiation of oriLyt-mediated DNA synthesis are likely to occur.

As previously shown, 11 loci are required to complement human cytomegalovirus (HCMV) DNA replication in a transient-transfection assay (G. S. Pari and D. G. Anders, J. Virol. 67:6979-6988, 1993). Six of these loci encode known or candidate replication fork proteins, as judged by sequence and biochemical similarities to herpes simplex virus homologs of known function; three encode known immediate early regulatory proteins (UL36-38, IRS1/TRS1, and the major immediate early region spanning UL122-123); and two encode early, nucleus-localized proteins of unknown functions (UL84 and UL112-113). We speculated that proteins of the latter five loci might cooperate to promote and regulate expression of the six replication fork proteins. To test this hypothesis we made luciferase reporter plasmids for each of the replication fork gene promoters and measured their activation by the candidate effectors, expressed under the control of their respective native promoters, using a transient-cooperativity assay in which the candidate effectors were subtracted individually from a transfection mixture containing all five loci. The combination of UL36-38, UL112-113, IRS1, or TRS1 and the major immediate early region produced as much as 100-fold-higher expression than the major immediate early region alone; omitting any one of these four loci from complementing mixtures produced a significant reduction in expression. In contrast, omitting UL84 had insignificant (less than twofold), promoter-dependent effects on reporter activity, and these data do not implicate UL84 in regulating HCMV early-gene expression. Most of the effector interactions showed significant positive cooperativity, producing synergistic enhancement of expression. Similar responses to these effectors were observed for the each of the promoters controlling expression of replication fork proteins. However, subtracting UL112-113 had little if any effect on expression by the UL112-113 promoter or by the simian virus 40 promoter-enhancer under the same conditions. Several lines of evidence argue that the cooperative interactions observed in our transient-transfection assays are important to viral replication in permissive cells. Therefore, the data suggest a model in which coordinate expression of multiple essential replication proteins during permissive infection is vitally dependent upon the cooperative regulatory interactions of proteins encoded by multiple loci and thus have broad implications for our understanding of HCMV biology.

Although the overall picture of HCMV DNA synthesis appears typical of the herpesviruses, some novel features are emerging. Six herpesvirus-group-common genes encode proteins that likely constitute the replication fork machinery, including a two-subunit DNA polymerase, a helicas-primase complex and a single-stranded DNA-binding protein. No homolog of the herpes simplex virus origin-binding helicase is evident, but at least one additional HCMV protein of unknown function, pUL84, appears to be required for initiation. Replication initiates within or near the large and structurally complex lytic-phase replicator, ori-Lyt, near the center of UL. Recent findings suggest that ori-Lyt-mediated initiation of DNA synthesis occurs through a mechanism distinct from that employed by herpes simplex virus.

The human cytomegalovirus (HCMV) DNA polymerase gene (UL54; also called pol) is a prototypical early gene in that expression is mandatory for viral DNA replication. Recently, we have identified the major regulatory element in the UL54 promoter responsive to the major immediate early (MIE) proteins (UL122 and UL123) (J. A. Kerry, M. A. Priddy, and R. M. Stenberg, J. Virol, 68:4167-4176, 1994), Mutation of this element, inverted repeat sequence 1 (IR1), abrogates binding of cellular proteins to the UL54 promoter and reduces promoter activity in response to viral proteins in transient-transfection assays. To extend our studies on the UL54 promoter, we aimed to examine the role of IR1 in UL54 regulation throughout the course of infection, These studies show that viral proteins in addition to the MIE proteins can activate the UL54 promoter, Proteins from UL112-113 and IRS1/TRS1, recently identified as essential loci for transient complementation of HCMV oriLyt-dependent DNA replication? were found to function as transactivators of the UL54 promoter in association with MIE proteins, UL112-113 enhanced UL54 promoter activation by MIE proteins three- to fourfold, Constitutive expression of UL112-113 demonstrated that the MIE protein dependence of UL112-113 transactivational activity was not related to activation of cognate promoter sequences, suggesting that UL112-113 proteins function in cooperation with the MIE proteins. Mutation of IR1 was found to abrogate stimulation of the UL54 promoter by UL112-113, suggesting that this element is also involved in UL112-113 stimulatory activity, These results demonstrate that additional viral proteins influence UL54 promoter expression in transient-transfection assays via the IR1 element, To confirm the biological relevance of IR1 in regulating UL54 promoter activity during viral infection, a recombinant virus construct containing the UL54 promoter with a mutated IR1 element regulating expression of the chloramphenicol acetyltransferase (CAT) reporter gene (RVIRmCAT) was generated, Analysis of RVIRmCAT revealed that mutation of IR1 dramatically reduces UL54 promoter activity at early times after infection, However, at late times after infection CAT expression by RVIRmCAT, as assessed by RNA and protein levels, was approximately equivalent to expression by wild-type RVpolCAT. These data demonstrate IR1-independent regulation of the UL54 promoter at late times after infection. Together these results show that multiple regulatory events affect UL54 promoter expression during the course of infection.

The genetically defined human cytomegalovirus (HCMV) lytic-phase replicator, oriLyt, comprises more than 2 kb in a structurally complex region that spans a variety of potential transcription control signals. Several transcripts originate within or cross oriLyt, and we are studying these oriLyt transcription units to determine whether they participate in initiating or regulating lytic-phase DNA synthesis. Results presented here establish the temporal accumulation and structure of the smallest replicator transcript, which we call SRT, and identify a single-sequence element essential to replicator function. SRT was detected as early as 2 h after HCMV infection of human fibroblast cells; transcript levels increased by 24 h and continued to increase thereafter. Consistent with its early appearance, treatment of HCMV-infected cells with the viral DNA polymerase inhibitor phosphonoformic acid had no effect on SRT accumulation; however, no SRT was detected in RNA preparations from cycloheximide-treated infected cells. Additional Northern (RNA) analysis localized the 0.2- to 0.25-kb SRT to an apparently noncoding segment near the center of the oriLyt core region. Reverse transcriptase PCR (rapid amplification of cDNA 5' ends [5'-RACE]) identified a single 5' end. In transient-transfection assays, the sequence immediately upstream of SRT functioned as a promoter responsive to HCMV infection when placed upstream of a reporter gene, suggesting that SRT is the product of a discrete transcription unit. RNA ligase-mediated 3'-RACE showed that SRT is not polyadenylated and has heterogeneous 3' ends within a roughly 45-nucleotide window overlapping an oligopyrimidine sequence having counterparts in the lytic-phase replicators of several herpesviruses. Mutation of the oligopyrimidine element showed that it is essential to oriLyt replicator function; it is the only essential single-sequence HCMV oriLyt replicator element described to date. Collectively, the location of SRT near the center of the oriLyt core region, its early expression, its overlapping relationship with a sequence element essential to replicator function, and its similarities to replicator transcripts in other systems suggest the possibility that SRT plays a role in initiating or regulating HCMV lytic-phase DNA synthesis.

Previous results showed that plasmids containing human cytomegalovirus (HCMV) oriLyt are replicated after transfection into permissive cells if essential trans-acting factors are supplied by HCMV infection (D. G. Anders, M. A. Kacica, G. S. Pari, and S. M. Punturieri, J. Virol. 66:3373-3384, 1992). We have now used oriLyt as a reporter of HCMV DNA replication in a transient complementation assay in which cotransfected cosmid clones, instead of HCMV infection, provided essential trans-acting factors. Complemented replication was oriLyt dependent and phosphonoformic acid sensitive and produced tandem arrays typical of HCMV lytic-phase DNA synthesis. Thus, this assay provides a valid genetic test to find previously unidentified genes that are essential for DNA synthesis and to corroborate functional predictions made by nucleotide sequence comparisons and biochemical analyses. Five cosmids were necessary and sufficient to produce origin-dependent DNA synthesis; all but one of these required cosmids contain at least one candidate homolog of herpes simplex virus type 1 replication genes. We further used the assay to define essential regions in two of the required cosmids, pCM1017 and pCM1052. Results presented show that UL44, proposed on the basis of biochemical evidence to be the HCMV DNA polymerase accessory protein, was required for complementation. In addition, three genomic regions encoding regulatory proteins also were needed to produce origin-dependent DNA synthesis in this assay: (i) IRS1/TRS1, which cooperates with the major immediate-early proteins to activate UL44 expression; (ii) UL36-38; and (iii) the major immediate-early region comprising IE1 and IE2. Combined, these results unequivocally establish the utility of this approach for mapping HCMV replication genes. Thus, it will now be possible to define the set of HCMV genes necessary and sufficient for initiating and performing lytic-phase DNA synthesis as well as to identify those virus genes needed for their expression in human fibroblasts.

Recently we described the use of human cytomegalovirus (HCMV) cosmid clones in a cotransfection assay of HCMV oriLyt replication (G. S. Pari, M. A. Kacica, and D. G. Anders, J. Virol. 67:2575-2582, 1993). We have now used this assay to identify 11 distinct required loci encoding trans-acting factors sufficient for transient complementation of oriLyt-dependent DNA replication. This set includes all of the virus genes essential to initiate and perform DNA synthesis together with the virus genes required to express these replication functions from their native promoters. Six of the identified loci span open reading frames (ORFs) that encode homologs or probable homologs of herpes simplex virus type 1 replication genes, consistent with predictions based on sequence similarities and biochemical properties. These include the DNA polymerase UL54 and polymerase-associated protein UL44, the single-stranded-DNA-binding protein UL57, and proposed subunits of a helicase-primase complex, UL70, UL105, and UL101-102. Frameshift mutations in any one of these essential ORFs abrogated complementation of DNA replication. Three required loci, UL36-38, IRS1 (or TRS1), and IE1/IE2, encode known regulatory proteins. The remaining two loci span ORFs UL84 and UL112-113 and encode early temporal class nucleus-associated proteins of unknown function. Neither of these genes have been implicated previously in DNA replication or in regulating gene expression, nor have counterparts in herpes simplex virus type 1 or Epstein-Barr virus been described. The results presented here will facilitate investigation of the mechanisms and regulation of HCMV lytic-phase DNA replication.