Identification of Positive and Negative Regulatory Regions Involved in Regulating Expression of the Human Cytomegalovirus UL94 Late Promoter: Role of IE2-86 and Cellular p53 in Mediating Negative Regulatory Function

Cell culture and virus.

Human embryonic lung (HEL) cells, human astrocytoma cell line U373(MG), and human osteosarcoma cell line Saos-2 were propagated in Dulbecco modified Eagle medium (DMEM) supplemented with 10% [HEL and U373(MG) cells] or 15% (Saos-2 cells) fetal bovine serum (FBS) as well as antibiotics. Towne strain HCMV [HCMV (Towne)] was propagated in HEL cells as previously described (40, 80).

DNA cloning, sequencing, and plasmids.

UL94 promoter fragments were cloned by PCR amplification of the appropriate regions by using Vent DNA polymerase (New England Biolabs) and a Perkin-Elmer Cetus DNA thermocycler. The 9412 region was generated by amplification from HindIII-digested HCMV (Towne) genomic DNA. The remaining constructs were amplified from plasmid pCAT9412. Oligonucleotides were synthesized at the UNC-Chapel Hill Lineberger Comprehensive Cancer Center Oligonucleotide Facility. All primers were engineered to contain either HindIII sites (upstream primers) or PstI sites (downstream primers) by inclusion of the following sequence at the 5′ end of each primer: (NNNN)₂AAGCTT for generation of HindIII sites or (NNNN)₂CTGCAG for generation of PstI sites, where the restriction site is underlined and NNNN corresponds to either ACGT (primers 1 and 2), CGAT (primers 3 to 8), or GCAT (primers 9 to 15). Each construct was designated by “94” followed by the numbers of the primer pair used in its amplification. The HCMV-specific region of each primer and the constructs generated using each primer are listed in Table 1.

Following amplification, fragments were digested with HindIII-PstI (Promega) and ligated into HindIII-PstI-digested pCATBasic (Promega). Ligations were transformed into Escherichia coli DH5α, and ampicillin-resistant clones were screened by colony hybridization using nick-translated, [³²P]dATP-labeled PCR fragments as described previously (76). Positive clones were grown, and plasmid DNA was purified by using Qiagen-tip 500 columns (Qiagen). Some promoter-chloramphenical acetyltransferase (CAT) constructs were analyzed by DNA sequence analysis using a Sequenase kit (United States Biochemical) as suggested by the manufacturer, while others were analyzed by fluorescence tagging at the UNC-Chapel Hill Automated DNA Sequencing Facility.

Expression plasmids for HCMV IE2-86 (pcDNA3-IE86) and IE1-72 (pcDNA-IE72) proteins have been previously described (80). Expression plasmids for wild-type (pC53-SN3) and mutant (pC53-SCX3) p53 were supplied by Bert Vogelstein and have also been previously described (5).

Primer extension analysis.

UL94 transcripts were analyzed by primer extension using whole-cell RNA isolated from HCMV-infected HEL cells as described previously (76). The first UL94-specific primer, UL94-3 (5′ CACCACGTCAGCGTACCAAGTCTGTTC 3′), used in these assays has also been previously described (76). The second primer, UL94-2, which overlaps the UL94 open reading frame (ORF), has the sequence 5′ ATGGCTTGGCGCAGCGGTAT 3′.

CAT assays.

For infection-transfection experiments, cells were seeded into 35-mm-diameter six-well plates at 3 × 10⁶ cells/well. The following day, cells were transfected via liposome-mediated transfection using 1,3-dioleoyloxy-2-(6-carboxyspermyl)propylamide (DOSPER; Boehringer Mannheim). For each 35-mm-diameter well, 0.5 μg of reporter plasmid along with 0.5 μg of Rous Sarcoma virus (RSV)–β-galactosidase (β-Gal) or simian virus 40 (SV40)–β-Gal plasmid was mixed with 4 μl of DOSPER in a final volume of 100 μl of HEPES-buffered saline (20 mM HEPES, 150 mM NaCl [pH 7.4]). Sixty microliters of the DNA-liposome complexes was added dropwise to cells cultured with 1 ml of medium. All transfections were done in triplicate and were allowed to proceed overnight. The next day, the transfectant was removed; cells were washed once with 2 ml of DMEM and subsequently infected with HCMV at a multiplicity of infection of approximately 2 to 5 PFU/cell. Following a 2-h absorption period, 1 ml of DMEM supplemented with 4% heat-inactivated FBS was added to each 35-mm-diameter well. For drug block experiments, the medium was supplemented with 10 μM ganciclovir [9-(1,3-dihydroxy-2-propoxymethyl)guanine (DHPG); Syntex] and was changed daily. Cells were harvested at the appropriate time point postinfection in 250 μl of 1× cell lysis buffer (Promega). For CAT assays, cell extract was mixed with acetylenzyme coenzyme A (Boehringer Mannheim) and [¹⁴C]chloramphenicol (New England Nuclear), and CAT assays were performed as described previously (83). Samples were standardized by using the Promega β-Gal enzyme assay system. Assays were carried out as suggested by the manufacturer, and absorbance at 420 nm for each sample was determined with a Beckman DU-70 spectrophotometer.

For cotransfection experiments, 0.1 to 0.5 μg of each effector plasmid was added to the transfection mixture along with 0.5 μg of each of the reporter and standardization plasmids. DNA amounts were standardized by inclusion of the appropriate amount of plasmid pGEM-7zf(+) (Promega). Transfections were carried out as described above except that following the transfection, the medium was replaced with 2 ml of DMEM supplemented with 10% (HEL and U373(MG) cells) or 15% (Saos-2 cells) FBS. At 72 h posttransfection, cells were harvested and CAT assays were performed as described above.

EMSA.

For p53 electrophoretic mobility shift assays (EMSA), we used purified baculovirus-expressed p53 protein with a six-histidine tag (57). Complementary oligonucleotides containing either wild-type or mutated p53-binding sites from the UL94 promoter were annealed to generate double-stranded probes. Sequences of oligonucleotides pairs (5′ to 3′) are as follows: 94p53W2, TCACGGAACATGTCCTGGCGC; 94p53C2, GCGCCAGGACATGTTCCGTGA; 94p53W3, TCACGGAACATGTCCTGGCGCGTTGTTTGGGAACTTTGCCGTCAT; 94p53C3, ATGACGGCAAAGTTCCCAAACAACGCGCCAGGACATGTTCCGTGA; 94p53m1, TCACGGAATCGCTCCTGGCGCGTTGTTTGGGAATCGCGCCGTCAT; and 94p53m2, ATGACGGCGCGATTCCCAAACAACGCGCCAGGAGCGATTCCGTGA.

EMSA were performed as previously described (40) except that 1× binding buffer consisted of 10% glycerol, 25 mM HEPES (pH 7.6), 50 mM NaCl, 1 mM dithiothreitol, 0.5 μg of bovine serum albumin/μl, 0.1% Triton X-100, and 0.1 μg of poly(dI-dC)/μl. For antibody supershift experiments, reactions were performed with 1 μl of antibody per 15 of μl reaction mixture for 30 min at room temperature prior to addition of the probe. Anti-p53 antibodies 421 and DO-1 were obtained from Calbiochem Oncogene Research Products.

Late-specific RNA start site usage in UL94 transcription.

We previously reported that UL94-specific DNA probes detected two classes of transcripts of approximately 9.1 and 2.0 kb in Northern blot analysis of HCMV-infected cell RNA (76, 77). Both transcript classes could be detected only at late times of infection and were sensitive to treatment with ganciclovir, suggesting that UL94 was a member of the true late kinetic class. We also mapped a putative RNA start site upstream of the UL94 ORF (76). This start site, located 336 nucleotides (nt) upstream of the UL94 initiation codon, was positioned 30 bp downstream of a TATA-box-like sequence and was detected by using RNA isolated from HCMV-infected cells at 72 h postinfection (hpi). To ascertain whether this start site was utilized exclusively at late times of infection, we performed primer extension analysis on RNA isolated from HCMV-infected cells at IE, early, and late times of infection, as well as RNA from mock-infected cells and HCMV-infected cells treated with ganciclovir, using a UL94-specific primer that was utilized in our previous RNA mapping experiments. As demonstrated in Fig. 1, extension of the primer to the putative RNA start site could be detected only at 72 hpi in the infection time course; in addition, no initiation of transcription at this site was detected at 72 hpi in the presence of ganciclovir or in mock-infected cells. Results consistent with those shown in Fig. 1 were also obtained in analyses using a second primer (UL94-2) which overlapped the UL94 ORF. These results confirm our previous observations that UL94-specific mRNA can be detected only at late times of infection and also demonstrate that the previously mapped UL94 transcript start site is utilized exclusively at late times of infection.

UL94 promoter sequence analysis.

To begin to understand how UL94 transcription is regulated, we sequenced a 525-bp region proximal to the TATA-box-like element located upstream of the UL94 ORF. The sequence analyzed included approximately 210 bp upstream of the UL94 RNA start site as well as 335 bp of sequence located between the RNA start site and the predicted UL94 initiation codon. The results (Fig. 2) are presented in comparison to the published sequence for HCMV strain Ad169 in this region (7). We observed only minor differences in nucleotide sequence between Towne and Ad169 in the putative UL94 promoter region. The likely TATA box at −30 is nonconsensus but is consistent in sequence with TATA-like elements found in other HCMV promoter regions. As in other HCMV early and late promoters, single consensus binding sites for several cellular transcription factors, including Sp1 and c-Myc, are located immediately upstream of the TATA box (15, 31). In addition, a possible 20-nt p53 binding element (with the two 10-nt half-sites separated by 13 nt of intervening sequence) is located in the upstream region. The upstream-most p53 half-site is a perfect match with the consensus p53-binding sequence WWWCATGRRR, while the downstream half-site has a 2-nt mismatch. The sequence and arrangement of these elements is consistent with that observed for cellular promoters which have been demonstrated to be regulated by p53 (18, 20).

Analysis of sequences located downstream of the UL94 RNA start site showed no additional TATA-like elements were found in this region, although we did detect several consensus initiator elements (YYAN[T/A]YY). We have thus far, however, been unable to detect transcription initiation at any of these sites, suggesting that the presence of these elements is completely fortuitous. No cis-regulatory sequence (crs)-like or other binding sites for the HCMV IE2-86 viral transactivator could be found in the UL94 promoter region. However, sequence elements for cellular transcription factors E2F, Sp1, CREB, and the TATA box have been demonstrated to mediate virus-induced transcription of viral and heterologous promoters, suggesting that these elements, or others, could play a role in regulation of UL94 transcription (41, 48, 74, 80).

Functional analysis of UL94 promoter sequences.

To examine the ability of the putative UL94 promoter sequences to direct transcription, we subcloned the 525-bp fragment sequenced as described above into reporter vector pCATBasic. Human fibroblasts were transfected with the resultant plasmid (9412CAT) and subsequently infected or mock infected with HCMV. As shown in Fig. 3A, extracts from 9412CAT-transfected fibroblasts contained significantly higher CAT activity following a 72 h superinfection with HCMV relative to mock-infected cells or infected cells which were transfected with the parent vector lacking promoter sequences. These results indicated that the 525-bp UL94 promoter fragment contains a functional promoter that is activated in a virus-dependent manner.

As a comparison to UL94 promoter activity in fibroblasts, we repeated the experiment with the HCMV permissive human astrocytoma cell line U373(MG) and obtained results analogous to the results obtained with fibroblasts (Fig. 3B) (13, 49, 56). It is worth noting that in quantitative comparisons between fibroblasts and U373(MG) cells, we generally observed higher levels of UL94 promoter activity in U373(MG) cells. A comparison of the transfection efficiencies of HEL and U373(MG) cells by in situ staining for β-Gal activity, following transfection with an RSV–β-Gal vector, indicated that U373(MG) cells were transfected at a significantly higher frequency than the HEL cells and with much less variability. Since U373(MG) cells are fully permissive for HCMV infection and are more efficiently transfected, we often used these cells preferentially in subsequent transfection-infection experiments. However, many experiments were also reproduced in HEL cells to ensure that the results were consistent within the context of normal human cells.

Deletion analysis of the UL94 promoter.

To more precisely define the important functional regions of the UL94 promoter, we generated a series of deletion mutants of the original 525-bp construct and analyzed the ability of these constructs to support expression of CAT in transfection-infection experiments. As can be seen in Fig. 4A, deletion of sequences between +130 and +335 resulted in no substantial loss in CAT activity as measured at 72 hpi in U373(MG) cells. In contrast, deletion of sequences between +2 and +130 resulted in at least a 60% reduction in CAT activity relative to the full-length construct. The relative decrease in activity observed between constructs 9414CAT and 9413CAT (i.e., 3-fold) is identical to the reported decrease in activity of HCMV DNA polymerase promoter-reporter constructs following deletion of the essential IR1 sequence element and analysis by transient transfection-infection (38). Thus, these results suggest that while sequences between +130 and +335 of the UL94 promoter region do not contribute significantly to the activity of the UL94 promoter, a putative positive regulatory region is located between +2 and +130.

To examine the influence of sequences upstream of the TATA box on UL94 promoter activity, we also created a series of 5′-deletion mutants, as shown in Fig. 4B. To minimize any confounding effects of the downstream positive regulatory region mapped above, we initially constructed a series of 5′-deletion constructs which extended only to the RNA start site and thus did not include the downstream regulatory region. As shown in Fig. 4B, when these constructs were analyzed by transfection-infection of U373(MG) cells, no significant CAT activity was observed in mock-infected cells; in addition, we observed a relatively low CAT activity from constructs containing 210 or 120 bp of upstream sequence. However, deletion of all sequences upstream of the TATA box (9137CAT) resulted in CAT activity at least threefold greater than for constructs which contained 120 bp or more of upstream sequence. The CAT activity from this minimal construct was similar to that observed with the 9414CAT construct, which contained 210 bp of upstream sequence as well as the putative downstream regulatory region. Since addition of 120 bp of upstream sequence resulted in diminished CAT activity relative to the minimal construct containing only the region immediately proximal to the TATA box and RNA start site, this experiment suggested that a possible negative regulatory region is located upstream of the TATA box in the UL94 promoter.

To demonstrate that the TATA sequence found in our promoter-CAT constructs was indeed a TATA box, we created a minimal promoter construct (94310CAT) in which specific point mutations were introduced into the TATA sequence (TATTATTAA to TAGGAGTAA). As shown in Fig. 4B, the level of CAT activity from this construct was threefold lower than that of the otherwise identical wild-type construct (compare 94310CAT and 9437CAT). This reduction in CAT activity is similar to that observed in constructs which contain a wild-type TATA box and the putative 5′ NRE, indicating that the wild-type TATA sequence is essential to maximum promoter activity.

To determine the effect of the putative 3′ PRE on promoter activity, with or without the presence of the 5′ NRE, we created an additional construct which contained no sequences upstream of the TATA box but extended 3 to +130 (i.e., into the PRE). As shown in Fig. 5, there was no significant difference in CAT activity between constructs 9447CAT and 9437CAT at 72 hpi. This finding suggested that the PRE region did not significantly add to the promoter activity that was observed with the TATA box alone at late times of infection. In contrast to these results, however, CAT activity from construct 9414CAT, containing the TATA box, PRE, and NRE, was significantly higher at 72 hpi than the activity of construct 9413CAT, which contains only the TATA box and NRE. These results suggested that the positive effect of the PRE is dominant over the negative effect of the NRE at late times of infection.

Taken together, the results of this series of 5′- and 3′-deletion constructs, as well as the TATA-box point mutants, suggests the existence of three important regulatory elements within a region encompassing −120 to +130 of the UL94 promoter: a 5′ NRE, a 3′ PRE, and a TATA box. At late times of infection, the repressive effect of the 5′ NRE is observed only when the 3′ PRE is absent, indicating that PRE function is dominant to NRE function at late times of infection. However, constructs which contain only the TATA box and the 5′ NRE exhibit only minimal activity in comparison to the TATA-only construct or constructs containing the 3′ PRE. Thus, the PRE appears to function as a derepression element, as it has the capacity to overcome the negative effect exerted by the NRE on the TATA box region of UL94 promoter at late times of infection.

UL94 promoter activity at 24 hpi and in the presence of ganciclovir.

True late viral promoters are active only after the onset of viral DNA replication. Deletion mutagenesis of late promoters from HSV has suggested that in addition to the requirement for viral DNA replication, cis-regulatory sequences located in late promoters influence promoter activity by either enhancing activity exclusively at late times of infection or by preventing late promoter activity at earlier times of infection (19, 21, 22, 27, 29, 30). Indeed, it is likely that both positive and negative regulatory mechanisms are necessary for precise regulation of late promoters. To address this question, we performed another series of transfection-infection experiments in which CAT activity was measured following a 24-h infection with HCMV or a 72-h infection in the presence of ganciclovir. Specifically, we compared the activity of our promoter construct containing only the TATA box (9437CAT) with those of constructs containing the TATA box and either the NRE (9413CAT), the PRE (9447CAT), or both (9414CAT) at both 24 hpi and at 72 hpi in the presence of ganciclovir. As can be seen in Fig. 5B, none of the UL94 promoter constructs had significant CAT activity at 24 hpi, as directly compared to the activity of the promoterless-vector control. There was a small but measurable increase in activity with constructs 9437CAT and 9447CAT, in comparison to the other constructs, suggesting that lack of the NRE resulted in higher activity of the UL94 promoter at early times of infection. This conclusion was substantiated by the results obtained at 72 hpi with ganciclovir: constructs 9437CAT and 9447CAT had significantly higher activity under these conditions than either 9413CAT or 9414CAT; indeed, the activity of 9413CAT and 9414CAT was no higher at 72 hpi with ganciclovir than at 24 hpi. Interestingly, whereas at 72 hpi the activity of 9414CAT was equivalent to that seen with 9447CAT or 9437CAT (Fig. 5A), its activity at 24 hpi and at 72 hpi with ganciclovir was extremely low. This suggests that the dominant effect of the PRE is a late-phase-specific event, since we do not observe the derepression at 24 hpi or at 72 hpi with ganciclovir. In addition, since both 9437CAT and 9447CAT, which lack the NRE, are active both at 24 hpi and at 72 hpi with ganciclovir, it is likely that the UL94 promoter TATA box can be activated at early times of infection in the absence of the NRE.

Additional deletion mapping of the 5′ NRE.

Sequence analysis of the UL94 promoter demonstrated that putative binding sites for several cellular transcription factors, including p53, were located within the NRE region (Fig. 2). It has previously been reported that whereas p53 generally activates transcription in a DNA-binding-dependent manner, in the presence of viral proteins such as HCMV IE2-86, adenovirus E1A or E1B, hepatitis B virus (HBV) X protein, or SV40 large T antigen, transcription from promoters with p53-binding sites is repressed (12, 28, 48, 55, 57, 62, 64, 71, 72). Since the NRE appears to be involved in repression of the UL94 promoter, we tested the effect of deletion of either one or both p53 half-sites on UL94 promoter activity in transfection-infection experiments. There was an additive effect of deletion of either one or both p53 sites: when both sites were deleted, there was an approximately 2.5-fold increase in UL94 promoter activity at 72 hpi in comparison to constructs containing the p53-binding-site region (Fig. 6A). The effect was even more pronounced at 72 hpi with ganciclovir: as much as a fivefold increase in UL94 promoter activity was seen when the region containing the p53 sites was deleted (Fig. 6B). In addition, we detected a close to twofold increase in promoter activity at late times of infection when point mutations were introduced into the upstream p53-binding site (Fig. 6C). These results suggest that the region containing the p53-binding sites contributes to the repression activity observed with the UL94 promoter NRE. Since deletion of the p53-binding sites did not restore promoter activity to that observed with deletion of the entire NRE, we conclude that other sequences besides the p53-binding-site region play a role in full NRE activity.

Effect of p53 on UL94 promoter activity in cotransfection assays.

To determine the effect of p53 on UL94 promoter activity, we cotransfected p53-negative Saos-2 cells with plasmids expressing either wild-type or mutant p53 and UL94 promoter construct 9439CAT, which contains the putative p53-binding sites. Cotransfection of 9439CAT with wild-type p53 resulted in an approximately 20-fold increase in promoter activity compared to cotransfection with mutant p53 or vector alone controls (Fig. 7A). Likewise, wild-type p53 had no effect on the promoterless vector control. In addition to promoter construct 9439CAT, we tested several other UL94 promoter constructs for responsiveness to p53 in Saos-2 cells (Fig. 7B). All of the constructs containing both p53-binding sites were significantly activated by cotransfection with wild-type p53, whereas constructs lacking the p53-binding sites were not activated. Interestingly, construct 9436CAT, which contains only the downstream, nonconsensus p53-binding site, and construct 94313CAT, which contains point mutations in the upstream p53-binding site, were also not activated by p53 in Saos-2 cells. These results imply that both p53-binding sites are necessary for responsiveness to p53 or, alternatively, that the downstream p53 site is simply unresponsive to p53. These two possibilities are addressed below.

Binding of p53 to the UL94 promoter.

To demonstrate binding of p53 to the UL94 promoter, we used, in EMSA, oligonucleotides containing either one or both UL94 promoter p53-binding sites, or oligonucleotides containing mutated p53 sites, and purified baculovirus-expressed p53. As shown in Fig. 8, p53 bound to the wild-type but not the mutated UL94 p53 sites (lanes 3 and 4). This binding could be specifically competed by excess unlabeled wild-type (W3C3) but not mutant (m1/m2) UL94 p53 site probe (data not shown). We could also supershift the p53-UL94 promoter complex with p53-specific antibodies (lanes 7 and 8). In addition, p53 protein bound at a much higher affinity to probes containing both p53 sites (lane 10 [with W3C3 probe]) than a probe containing only the upstream consensus p53-binding site (lane 9 [with W2C2]). This result suggests that binding of p53 to the UL94 promoter requires both p53-binding sites and is consistent with previous reports which indicate that binding of tetrameric p53 to responsive promoters requires more than one 10-bp p53 consensus sequence. These results also suggest that the downstream p53 site in the UL94 promoter is likely to be necessary for stabilization of the p53-DNA interaction, despite the fact that this element alone cannot support p53-mediated transactivation. It should be noted that while the gel shift and CAT assay data demonstrate that the two 10-bp p53-binding sites in the NRE are essential to obtain p53 binding to the NRE (EMSA) and transactivation (p53 expression alone) or suppression (coexpression of p53 and IE2-86 [see below]) of NRE-containing promoters (CAT assays), we have not yet determined which NRE nucleotides are bound to p53.

Effect of HCMV IE2-86 on the UL94 promoter.

Since the p53-binding sites in the UL94 promoter appear to contribute to repression of the UL94 promoter during infection, we wondered why they mediated p53-dependent activation in Saos-2 cells. As mentioned earlier, the HCMV IE2-86 protein has been reported to tether a transcriptional repression domain to p53 through protein-protein interaction (71). The effect of this interaction is a p53-binding-site-dependent IE2-mediated repression of transcription by p53. To examine the effect of IE2-86 on p53-mediated activation of the UL94 promoter, we cotransfected Saos-2 cells with UL94 promoter construct 9439CAT and expression plasmids for both p53 and IE2. As can be seen in Fig. 9A, IE2-86 alone had no effect on UL94 promoter activity; in contrast, when cotransfected with p53, IE2-86 abrogated p53-mediated activation of the UL94 promoter. The level of repression that we observed is consistent with previously published data showing repression of p53-mediated transcription by IE2-86. Furthermore, Western blot analysis showed this repression was not due to a decrease in p53 levels in the presence of IE2-86 (data not shown). These results demonstrate that IE2-86-mediated repression of p53 can recapitulate the transcriptional repression observed during infection by the UL94 promoter NRE region.

Of the functionally important UL94 promoter regions, construct 9439CAT contains only the TATA box and NRE; however, we also wanted to measure the effect of p53 on promoter constructs containing the PRE region, since our transfection-infection data suggested that the PRE had the capacity to disrupt NRE function at late times of infection. In addition, we were interested in more precisely localizing the PRE region. To address the second point, we generated an additional deletion mutant, 94115CAT, which extends to +48 with respect to the RNA initiation site. As can be seen in Fig. 9B, when this construct was tested for functionality in infection-transfection assays, its activity was similar to that of 9414CAT, which contained an additional 82 bp of downstream sequence. We conclude from this experiment that PRE function resides in the first 48 bp downstream of the RNA initiation site. To determine the effect of p53 and IE2 on 94115CAT, we cotransfected Saos-2 cells with 94115CAT and expression plasmids for p53 and IE2-86, as before. The results (Fig. 9C) indicated that 94115CAT is also responsive to p53 and, further, that IE2-86 could abrogate this stimulation. Thus, in the absence of any other viral gene products, the PRE region of UL94 promoter has no obvious role in the repression of the UL94 promoter activity by IE2-86 or p53.