The Initiator Element in a Herpes Simplex Virus Type 1 Late-Gene Promoter Enhances Activation by ICP4, Resulting in Abundant Late-Gene Expression

Abstract

The start site regions of late genes of herpes simplex virus type 1 are similar to the eukaryotic initiator sequence (Inr), have been shown to affect the levels of expression, and may also play a role in transcription activation by the viral activator ICP4. A series of linker-scanning mutations spanning the start site of transcription and several downstream mutations in the true late gC promoter were analyzed in reconstituted in vitro transcription reactions with and without ICP4, as well as in the context of the viral genome during infection. The nucleotide contacts previously found to be important for Inr function were also found to be important for optimal induction by ICP4. While the Inr had a substantial effect on the accumulation of gC RNA during infection, no other sequence downstream of the TATA box to +124 had a significant effect on levels of expression during infection. Therefore, these studies suggest that TATA box and the Inr are the only cis-acting elements required to achieve optimal expression of gC, and that the high levels of late-gene transcription may be largely due to the induction by ICP4, functioning through the Inr element.

During productive infection by herpes simplex virus type 1 (HSV-1), approximately 80 genes are sequentially expressed in three major kinetic groups: immediate-early (IE, α), early (E, β), and late (L, γ) genes (36, 37, 46, 60, 61). Each HSV-1 gene has its own promoter with an obvious TATA box homology that is recognized by the host RNA polymerase II transcription machinery (1, 11, 89). A viral protein VP16, carried in with the virion, activates transcription of the five IE genes in a complex with two cellular proteins, Oct-1 and HCF (5, 55, 66). Once expressed, the IE gene products regulate expression of the early and late genes by a variety of mechanisms in the course of infection (75).

One of the IE proteins, ICP4 (infected-cell polypeptide 4), is essential for viral growth as it is required for optimal expression of most early and late genes (15, 19, 22, 25, 67, 68). ICP4 is a nuclear phosphoprotein that acts as a homodimer to activate or repress transcription depending upon the promoter (12, 15, 26, 54, 63, 67, 71). Genetic and biochemical studies have shown that ICP4 binds to DNA and interacts with a component(s) of general RNA polymerase II transcription machinery to activate or repress transcription (2, 8, 17, 29, 42, 64, 65, 84). ICP4 forms a tripartite complex with TFIIB and TATA binding protein (TBP) on DNA, interacts with TBP-associated factor 250 (8, 84), and promotes the formation of transcription preinitiation complexes on promoters (28).

The transcriptional regulation of HSV genes during infection is mostly determined by the structure of a given promoter. The promoters of IE and E genes contain a number of cis-acting elements upstream of the TATA box (89, 90). Viral and cellular proteins (Oct-1/VP16, Sp1, CTF or NF1, etc.) bind to these cis-acting elements to activate transcription (21, 24, 25, 45, 89). An important characteristic of late genes is that their expression is significantly influenced by viral DNA replication. Depending on their DNA replication requirement for expression, late genes are further subdivided into leaky-late (βγ, γ₁) and true-late (γ, γ₂) genes. Expression of true-late genes is strictly dependent on viral DNA replication (10, 34, 36, 37, 46). The mechanism underlying the requirement for DNA synthesis has yet to be determined (53, 76, 82). In contrast to IE and E gene promoters, true-late gene promoters lack upstream cis-acting elements, and the region downstream of the TATA box may play a more important role in regulating their transcription and temporal expression (23, 32, 35, 38, 44, 50, 70, 79, 85).

The TATA box is common to all HSV genes and plays a major role in determining the absolute level of expression. However, the sequence of the TATA box itself does not greatly affect the kinetics of expression (36, 37, 46, 60, 61). The core promoters of several late genes also possess a sequence resembling a eukaryotic initiator (Inr) element at the start site region (32, 39, 78, 86). The Inr can be found on various cellular and viral promoters (52, 92) and has a loose consensus sequence, YYA₊₁N(T/A)YY (where Y is pyrimidine and N is any nucleotide). It can direct the specific initiation of transcription by itself or synergistically in conjunction with other promoter elements (3, 20, 57, 69, 83, 92). While the activity of the late core promoter elements has been examined (32, 39, 78, 85), the specific roles of the start site region or Inr elements have not been studied in detail. In addition to the promoter, it has been reported that the leader region is also involved in the expression of late genes. A cis-acting element called downstream activation sequence (DAS) in the leader region of U_L38 was found to be important for U_L38 expression (31, 32).

The HSV-1 glycoprotein C sequence (gC, or U_L44) is a late gene whose promoter contains a fairly strong TATA box centered at −25 relative to the start site of transcription (35). Transcription of gC is activated by ICP4, which facilitates the formation of the transcription initiation complex on the promoter (26, 27, 30, 79). Mutations in the core promoter region decrease the expression of gC (28, 85). Like other late genes, the sequence at the start site of gC transcription is similar to the consensus Inr sequences (28, 85). When the thymidine kinase gene (tk) start site region was replaced by that of gC, this chimeric promoter was activated by ICP4 better than wild-type (wt) tk promoter (28). Furthermore, while a mutation at the −1 site decreased transcription in the absence and presence of ICP4, transcription in the presence of ICP4 was affected to a greater degree (28). These results suggest that activation of the gC promoter by ICP4 may be mediated in part through the Inr. The presence of an Inr sequence in several late promoters (32, 39, 78, 85) also raises the possibility that the Inr is involved in the temporal regulation of their expression.

In this study, a series of linker-scanning (LS) mutations was introduced into the gC promoter region around the start site. The effects of the mutations on basal and ICP4-activated transcription were studied both in vitro and in vivo in the tk locus. Expression levels at different times after infection and the effect of DNA replication on expression were also determined. We also investigated contribution of the gC leader and start site regions at the natural gC locus during viral infection. The results suggest that (i) the gC start site contains a functional Inr, (ii) the same nucleotide contacts that are important for Inr function are also important for ICP4-activated transcription, (iii) Inr function is crucial for the accumulation of high levels of gC mRNA late after infection, and (iv) the TATA box and the Inr are the only cis-acting elements required for full expression of gC during viral infection.

MATERIALS AND METHODS

Viruses and cells.

wt (KOS) and the ICP4-deficient n12 viruses were propagated and their titers were determined on Vero cells and on the ICP4-complementing cell line E5, respectively, as previously described (14). The gCLS/KOS, gC-gfp, gC-Inr*, gC-DPE*, gC-DAS*, and gCΔ+14/+118 viruses were grown on Vero cells. The gCLS/n12 viruses were grown on E5 cells. The n12 mutant contains a nonsense mutation that truncates ICP4 at amino acid 251 (16).

Plasmid constructions.

Two parental plasmids, pgCHB (41) and pLSWT (28), were used to construct the series of LS pgCLS mutants. The LS mutations were introduced into the gC promoter region spanning −17 to +13 relative to the gC transcription initiation site by introducing new SacI restriction sites using oligonucleotides. The mutagenic oligonucleotides contained the gC sequences from −22 (BspEI) to +14 (BglII) with the SacI site at different positions (Fig. 1B). The oligonucleotides were inserted into pgCHB, replacing sequence −22 (BspEI) to +124 (BglII). The HindIII-BglII fragments from these plasmids were isolated and inserted into pLSWT, replacing the tk sequence from −500 (HindIII) to +55 (BglII). The resulting series of pgCLS plasmids contain the tk sequence from −500 to +2844 with the tk promoter sequence from −111 to +55 replaced by the gC sequence from −35 to +13. These plasmids also contain a temperature-sensitive (ts) mutation in the tk coding region (9, 40). All of the plasmids were sequenced to confirm the integrity of mutation.

The plasmids pSXgC, pUC18gC, pUC20gC, and pSXgC-gfp were used to generate the gC-gfp, gC-Inr*, gC-DPE*, and gC-DAS* viruses. pSXgC was constructed by inserting the SacI-XbaI fragment containing the gC region (from −2057 to +1495) from pXbaE (courtesy of C. Smith, University of Pittsburgh) into pGEM-Zf11(+) (Promega). A PstI-EcoRI fragment of pSXgC was subcloned into pUC18 to generate pUC18gC. pUC20gC was generated by subcloning the HincII-NruI fragment of pUC18gC into pUC20. pUC20 is derived from pUC18 by inserting the oligonucleotides 5"-CGAGCTAGCACGCGTCGCG-3" and 5"-AATTCGCGACGCGTGCTAGCTCGAGCT-3" into the SacI site of pUC18. To construct pSXgC-gfp, an AseI-MluI fragment containing the green fluorescent protein (GFP) gene driven by the HCMV promoter from pGFPX (77) was isolated, modified with EcoRI linkers (New England Biolabs), and inserted into the EcoRI site of pSXgC. pSXgC-Inr*, pUC18gC-Inr*, and pUC20gC-Inr* have the same sequence as pSXgC, pUC18gC, and pUC20gC, respectively, except that they possess the LS8 SacI mutation (Fig. 1B) at the gC start site. The SacI mutation (LS8) was introduced into pUC20gC to generate pUC20gC-Inr* by PCR site-directed mutagenesis using the mutagenic primers 5"-GGCCTACCCGAGCTCCCGAGGGGG-3" and 5"-CGCCCGCGGGAGCTCGGGTAGCCC-3" and the M13/pUC sequencing primers (New England Biolabs). Using 50 ng of pUC20gC as a template, a PCR was carried out in the presence of 25 mM KCl, 10 mM Tris-HCl (pH 8.83), 5 mM (NH₄)₂SO₄, 2 mM MgSO₄, 0.2 mM deoxynucleoside triphosphates, 15 pmol of each primer, and 1.25 U of high-fidelity Pwo DNA polymerase (Boehringer Mannheim Biochemicals) in a final volume of 50 μl. The PCR ran for 25 cycles of 30 s at 94°C, 30 s at 45°C, and 45 s at 72°C (Perkin-Elmer Gene Amp PCR system 2400). The resulting PCR products were digested with HincII-NheI and substituted in place of the corresponding fragment in pUC18gC. The integrity of the mutation was confirmed by sequencing the plasmid in both directions (with an ABI Prism sequencer). Plasmids pSXgC-DPE* and pSXgC-DAS* were also constructed in the same way as pSXgC-Inr* except that different sets of mutagenic primers were used and the PCR products for subcloning were digested with PstI and NheI. To construct pSXgC-DPE*, the oligonucleotide primers 5"-GAGGGCGCTGATCAAGGCCG-3" and 5"-CGGCCTTGATCACAAGCGCCCTC-3" were used. The primers 5"-CGCTTGGTCGGTTAACCGCATCG-3" and 5"-CGATGCGGTTAACCGACCAAGCG-3" were used to construct pSXgC-DAS*.

FIG. 1.

The initiator sequence (Inr) and construction of LS mutations at the gC start site. (A) Comparison of the start site sequences of several HSV-1 late genes, including gC, with other Inr elements and the Inr consensus sequence. The major transcription initiation sites are underlined. The stars represent the nucleotides most deleterious to Inr activity when singly mutated (43, 57). (B) Schematic representation of the templates (pgCLS plasmids) used in the in vitro transcription assay. The pgCLS plasmids contain the gC promoter region from −34 to +14 (shaded box) replacing the tk promoter region from −111 to +55. The gC TATA box (white box) and the SacI restriction sites specifying the LS mutations (black box) are shown. The line represents the approximately 3 kb of tk DNA sequences (between −500 and + 2860) flanking the gC promoter. The mutated gC region from −16 to +14 is also shown in more detail, with the new SacI restriction sites underlined. The mutated nucleotides are in outlined letters, and the major start sites of gC are indicated by stars. The restriction enzyme cleavage sites shown are as follows: S, SacI; H, HindIII; B, BamHI; Bg, BglII.

The plasmid pSXgCΔ+14/+118 was constructed to generate gCΔ+14/+118 virus. An EcoRI-PstI fragment of pSXgC plasmid was replaced by an EcoRI-PstI fragment of pUC18gCΔ+14/+118. The pUC18gCΔ+14/+118 was derived from pUC18gC by replacing BspEII-NruI (−25 to +118) with the following synthetic oligonucleotides: 5"-CCGGAAGGGGACACGGGCTACCCTCACTACCGAGGGCG-3" and 5"-CGCCCTCGGTAGTGAGGGTAGCCCGTGTCCCCTT-3".

Nuclear extract and ICP4.

The nuclear extract was prepared from HeLa cells as described previously (18, 30, 56, 74). ICP4 was purified by sequential column fractionation of infected Vero cell nuclei as previously described (80, 81).

In vitro transcription and primer extension.

Supercoiled DNA templates (50 ng) were incubated in the presence or absence of ICP4 and nuclear extract (6 μl). The final concentrations of components in reaction buffer were 40 mM HEPES-KOH (pH 7.9), 60 mM KCl, 12% glycerol, 8.3 mM MgCl₂, 0.6 mM ribonucleoside triphosphates, 0.3 mM dithiothreitol, and 12 U of RNasin (Promega). The reaction was carried out in final volume of 30 μl for 1 h at 30°C and stopped by adding 70 μl of stop solution (150 mM sodium acetate [pH 5.3], 15 mM EDTA [pH 8.0], 150 μg/ml of tRNA). The reaction mixture was then phenol extracted, and the transcription products were ethanol-precipitated.

For the primer extension analysis, the RNA synthesized in vitro was annealed to 3 ng of ³²P-labeled primer 5"-AGGGGTACGAAGCCATACGCGCTTCTACAAGGCGCT-3" complementary to tk sequence from +90 to +125, with a buffer containing 10 mM Tris-HCl (pH 7.5), 250 mM KCl, and 1 mM EDTA (pH 8.0). Annealing was carried out in a final reaction volume of 10 μl. The annealed products were reverse transcribed by using 300 U of Moloney murine leukemia virus reverse transcriptase (Gibco/BRL) in a reaction mixture containing 50 mM Tris-HCl (pH 7.5), 75 mM KCl, 3 mM MgCl₂, 10 mM dithiothreitol, 1 mM deoxynucleoside triphosphates, 12 U of RNasin, and 50 μg of actinomycin D. Reverse transcription was carried out in a final volume of 40 μl for 1 h at 42°C and stopped by addition of 60 μl of stop solution (1 M ammonium acetate, 20 mM EDTA). After phenol extraction and ethanol precipitation, the final products were analyzed on 6% denaturing polyacrylamide gels, dried, and exposed to XAR film (Kodak) for autoradiography. The radioactive signals were also quantified using a radioanalytic imaging system (Ambis Inc., San Diego, Calif.) or a PhosphorImager (Storm 840; Molecular Dynamics).

Primer extension reactions with RNA isolated from infected cells were carried out using the same conditions as above except that 6 μg of isolated total RNA was used in each reaction mixture. The same primer was used to detect the tk message and the primer, 5"-GTTCCCGGCAAAGCGAGACCGGGCATGAAAAC-3", complementary to the gC region from +70 to +102 was used to detect the gC message. In the gCΔ+15/+118 experiment, an oligonucleotide spanning from +170 to +200, 5"-GGACACCCCCGCCCCGAGCCACAACAGGCTCC-3", was used to detect gC message. As an internal control for some of the experiments, the primer 5"-CAACCCCCCGGGTAACCACGGGGTGC-3", complementary to the ICP8 sequence from +81 to +106, was also used.

Recombinant viruses.

All the gCLS/KOS mutants that carry the gCLS mutations in the tk locus of KOS were generated from the corresponding pgCLS plasmid by cotransfection of 10⁶ Vero cells with 3 μg of KOS viral DNA and 3 μg of linearized pgCLS plasmid as previously described (14). The recombinant viruses, which have a ts Tk phenotype, were isolated in the presence of 120 mM acyclovir (Sigma) at 39.5°C. Acyclovir-resistant plaques were amplified and screened by Southern blot analysis. To confirm the transfer of LS mutations to the tk locus, SacI-digested viral DNAs were fractionated on agarose gels, transferred to nitrocellulose membranes, and hybridized to ³²P-labeled pLSWT (28).

The gCLS/n12 mutants carrying both the LS mutations and a nonsense mutation in ICP4 were generated by crossing the n12 and corresponding gCLS/KOS viruses. 10⁶ E5 cells were coinfected with the gCLS/KOS and n12 viruses (multiplicity of infection [MOI] = 10), and the resulting recombinants were isolated by selecting for the ts Tk phenotype. The LS mutation was confirmed as described above. To confirm the presence of the ICP4 nonsense mutations, viral DNA was isolated, digested with BamHI and HpaI, and used for Southern blot analysis using ³²P-labeled BamHI-Y fragment from wt ICP4 as a probe (14).

The gC-gfp virus was generated by cotransfecting Vero cells with SacI-linearized pSXgC-gfp and KOS DNA. Green fluorescent plaques were isolated (Nikon FXA photomicroscope) and plaque purified twice. The insertion of the GFP gene into the gC coding region was confirmed by Southern blot analysis. The gC-Inr*, gC-DPE*, gC-DAS*, and gCΔ+14/+118 viruses were generated by cotransfecting XbaI-linearized pSXgC-Inr*, pSXgC-DPE*, pSXgC-DAS*, and pSXgCΔ+14/+118 with the gC-gfp viral DNA into Vero cells, respectively. Nonfluorescent plaques were isolated and plaque purified. The introduction of the mutation and loss of the GFP gene were confirmed by Southern blot analysis.

Infected-cell RNA and Northern blots.

To isolate total RNA, 5 × 10⁶ Vero or E5 cells were infected at an MOI of 10. Total cellular RNA was prepared from the infected cells using Ultraspec (Biotecx Laboratories, Inc.) at the indicated time after infection. For experiments done in the presence of phosphonoacetic acid (PAA) (Lancaster Synthesis, Eastgate, England), the medium used during and after infection was supplemented with PAA (400 μg/ml).

For Northern blot analysis, the indicated amount of total RNA was separated on a 1.3% agarose gel containing formaldehyde and transferred to nitrocellulose membrane as previously described (40). The blots were initially hybridized to a probe for the tk gene, then stripped and hybridized to a probe for ICP8. The nick-translated probes used included the SacI-SmaI fragment (+555 to +1217) for tk and the SalI-BamHI fragment (+585 to +1481) for ICP8. The blots were exposed for autoradiography, and the radioactive signals were quantified as described above.

RESULTS

The start site regions of many late genes, including the gC promoter, have similarities to the consensus initiator (Inr) sequence (26, 39, 78, 85, 86) (Fig. 1A) and affect transcriptional activity (26, 39, 78, 85). Mutational analyses on initiators in basal gene expression systems have shown that single changes in the +1 and +3 positions have the greatest effect on Inr activity (43, 57). We have previously found that the sequence at the start site of gC transcription affects the ability of ICP4 to activate transcription of the promoter (28). One explanation for these results is that ICP4 helps TFIID to utilize the Inr. A genetic test of this hypothesis would be to determine if the same changes that affect Inr activity also affect the magnitude of ICP4 induction.

To address this, a series of plasmids containing LS mutations around the gC start site region of the core promoter (−35 to +14) was constructed and substituted in place of the tk promoter in a plasmid containing the tk gene. The site of the LS mutation is marked by a new SacI restriction site (Fig. 1B).

In vitro analysis of LS mutants.

These plasmids containing the LS mutations were used directly in in vitro transcription reactions in the presence and absence of ICP4. The plasmids pgCLS1- to -5, -11, and -12 had no significant effect on transcription in vitro with or without ICP4 (data not shown; Fig. 2). The transcription of templates pgCLS6 to -10 (LS6 to LS10) and two unaffected templates, pgCLS1 and -11 (LS1 and LS11), were investigated in more detail. Consistent with previous results (28), ICP4 activated transcription from the wt gC promoter in a concentration-dependent manner (Fig. 2A). A number of the templates containing mutations around the start site resulted in reduced basal and ICP4-activated transcription. For example, relative to LS1, transcription from LS8 was reduced in the absence of ICP4, but was more impaired in the presence of ICP4, resulting in a lower fold induction (Fig. 2A). Additionally, the start site changes were evident using the template LS8.

FIG. 2.

In vitro transcription of the LS templates. (A) A hundred nanograms of each template (pgCLS1 or pgCLS8) was added to the in vitro transcription reactions with increasing amounts of ICP4. The primer used in this assay is complementary to the tk region from +90 to +125 (Fig. 1B). The final products were separated on a denaturing 6% polyacrylamide gel. For each set of reactions for each template, the first lane represents transcription in the absence of ICP4. The second, third, and fourth lanes are products in the presence of 50, 100, and 200 ng of ICP4, respectively. (B) Quantitative representation of induction ratio by ICP4 with each template. Fold induction of each template was normalized with fold induction of pgCLS1 at the different concentrations of ICP4 as defined in the text.

The experiment shown in Fig. 2A was independently conducted using templates LS6 to -10 and LS11. Therefore, each template was separately compared to LS1. The fold induction by ICP4 of each template, and at each concentration of ICP4, was normalized to the fold induction of LS1 by ICP4 at the corresponding ICP4 concentration (Fig. 2B). This induction ratio is a measure of how well ICP4 can activate transcription of a given template relative to LS1. The activation of LS6 and -11 by ICP4 were not significantly affected relative to LS1. By contrast, the ability of ICP4 to activate LS7 to -10 was significantly reduced. Therefore, the most affected templates (LS7, -8, -9, and -10) with respect to ICP4 activation are those containing mutations affecting the contacts previously shown to be the most important for Inr function (43, 57).

Analysis of Inr and leader mutants in virus infection.

To assess the contribution of this region to ICP4-activated transcription levels in the context of viral infection, recombinant viruses were generated containing the pgCLS1, pgCLS6 to -10, and pgCLS11 constructs in the tk locus of wt (strain KOS) virus. The expression of tk from these viruses again demonstrated that the pgCLS7, -8, -9, and -10 viruses were the most impaired, and with the possible exception of pgCLS8 all were expressed to a higher degree at early compared to late times postinfection (p.i.) (Fig. 3A).The addition of PAA did not eliminate expression from these constructs (data not shown). Similar results were obtained by analyzing the abundance of RNA by primer extension (Fig. 3B). Moreover, the primer extension products of LS7, -8, and -9 noticeably differed relative to the other constructs, indicating a change in start site. The common change in these constructs that is not present in the others is a change from the invariant A residue at position +1.

FIG. 3.

Expression kinetics of the gCLS/KOS viruses and the effect of the LS mutations on the transcription start site during virus infection. (A) Primer extension analysis of RNA isolated at different times p.i. from Vero cells infected (MOI = 10) with KOS and gCLS/KOS1, -6, -7, -8, -9, -10, and -11 viruses. The reactions were carried out in the presence of both the tk and ICP8 primers. ICP8 was used as a control, and products were separated on a denaturing 6% polyacrylamide gel. The virus designations are shown at the top. (B) Vero cells were infected (MOI = 10) with the gCLS/KOS viruses, and total RNA was isolated at 6 h p.i. The first four lanes represent a sequencing ladder of the gC start site region using the same primer that was used in the primer extension reactions. The major start sites are marked with asterisks and stars. The gCLS/KOS virus designations are shown at the bottom.

To ascertain if the reduction in expression in virus infection may in part be due to a reduction in the ability of ICP4 to activate Inr promoter mutants by ICP4, the viruses gCLS1, -6, -8, and -9 were genetically crossed with the virus n12 to put these promoters in the n12 background. Because n12 does not express functional ICP4 (16), the expression of the tk gene in Vero cells reflects expression in the absence of ICP4, and infection of E5 cells reflects expression in the presence of ICP4. Accordingly, Vero and E5 cells were infected with the gCLS/n12 recombinant viruses, and total RNA was isolated at 4 h p.i., which was subsequently analyzed by Northern blotting (Fig. 4A). The levels of ICP8 mRNA in the samples were determined to serve as a loading control. As expected, the levels of tk mRNA in gCLS/n12-infected Vero cells were very low in the absence of ICP4, but the differences in expression levels among the Inr mutants could still be discerned. In E5 cells, the most noticeably impaired promoter was the LS8 mutation. To quantify these signals, the amount of tk was normalized to the amount of ICP8 for a given cell type. The fact that eightfold more RNA prepared from Vero cells was used in the analysis was also considered. From these considerations it was calculated that there was about 400-fold more tk RNA in gCLS1-infected cells in the presence of ICP4 than in its absence (Fig. 4B). In contrast, the abundance of tk RNA in gCLS8-infected cells was only 100-fold greater in the presence of ICP4 than in its absence. The amount of induction by ICP4 in gCLS9-infected cells was also reduced. Therefore, consistent with the results of the in vitro data, ICP4 was less efficient in inducing the Inr-mutated promoters.

FIG. 4.

Effect of the start site region on ICP4-activated transcription in virus infection. (A) Northern blot analysis of tk expression in Vero and E5 cells infected at a MOI of 10 PFU/cell. Total RNA (5 μg from E5 cells and 40 μg from Vero cells) from infected cells was separated on a formaldehyde gel. The blot was first probed for tk and then stripped and reprobed for ICP8. (B) Quantitative representation of results shown in panel A. The signals from the blot were quantified and normalized to the ICP8 signal, and ratios of fold activation in the presence (E5) and absence (Vero) of ICO₄ were calculated. Each value represents the mean and standard deviation (error bars) from three independent experiments.

To more directly address the contribution of the Inr and putative cis-acting sites downstream of the Inr to gC expression during infection, four mutants were constructed in the gC locus of KOS. One of the mutants, gC-Inr*, contains the LS8 mutation and is otherwise intact with respect to the remainder of the gC locus. The viruses gC-DPE* and gC-DAS* also have the intact gC locus but have specific sequences changed in the leader region. In gC-DPE*, the sequence between +17 and + 23 was changed from GGTCGGG to TGATCAG (DPE consensus, [G/A]G[T/A]CGTG). In gC-DAS* the sequence between +22 and +28 was changed from GGAGGCC to GTTAACC. This corresponds to a change in the previously described element, which was found to partially substitute for a DAS element (31). The final mutant, gCΔ+15/+118, deletes the region between +15 and +118 of the gC leader at the gC locus.

Vero cells were infected with the four mutants and wt virus in the presence and absence of PAA, and RNA was extracted at different times p.i. Primer extension analysis was carried out simultaneously using primers for an early gene (tk) and for gC (Fig. 5). A different gC primer had to be used to compare KOS and gCΔ+15/+118 (Fig. 5B), because of differences in the leader regions. The results of the experiment using the gC-Inr* mutant (Fig. 5A and 6) reveal that (i) the start site of gC transcription from gC-Inr* was affected as anticipated from previous results, (ii) the kinetics of gC transcription was not significantly affected by the gC-Inr* mutation, and (iii) the levels of gC transcription were dramatically reduced as a function of the gC-Inr* mutation. Figure 6 shows a quantitative representation of the data in Fig. 5A for the internal controls (tk) and the gC genes expressed in KOS and gC-Inr*-infected cells. It is clear that the Inr sequence of gC has a profound effect on the accumulation of this abundantly transcribed late message.

FIG. 5.

Effect of the initiator and the leader region mutations at the gC locus. (A) Primer extension analysis of RNA isolated from Vero cells infected (MOI = 10) with KOS, gC-Inr*, gC-DPE*, and gC-DAS* mutants. RNA isolated from the indicated viruses at different time points was used in primer extension analysis in the presence of both the tk and gC primers. To detect the gC message, an oligonucleotide spanning the sequence from +70 to +102 was used as the primer. To detect tk message, the same primer described in Fig. 2 was used. (B) Primer extension analysis of RNA isolated from Vero cells infected with gCΔ+15/+118 mutant in the presence (+) and absence (−) of PAA. To detect gC message in this experiment, an oligonucleotide spanning the sequence from +170 to +200 was used.

FIG. 6.

Effect of Inr mutation in vivo. Transcription levels of gC from the wt and gC-Inr* mutant from Fig. 5A were quantitatively compared using a phosphorimager.

In contrast to the effects seen with the mutant initiator, the effects seen in gC-DPE*, gC-DAS*, and gCΔ+15/+118-infected cells were relatively minor, if present at all. While the level of gC RNA in gC-DAS*-infected cells may be slightly reduced relative to that in KOS-infected cells, this reduction is not significant relative to that seen in gC-Inr*-infected cells (Fig. 5A and 6). Furthermore, deletion of the entire leader between +15 and +118 had no significant effect on the accumulation of gC mRNA (Fig. 5B). Lastly, the accumulation of gC mRNA in all the infections was completely dependent on DNA synthesis (data not shown and Fig. 5B). Thus, it appears that the sequence having the greatest effect on the accumulation of gC mRNA is the Inr region and that downstream sequences to +118 have little effect on gC gene expression.

DISCUSSION

gC start site region as an Inr and facilitator of ICP4 activation.

Mutagenesis of the terminal deoxynucleotidyl transferase (TdT) and adenovirus major late (AdML) Inrs has shown that Inr activity cannot be abrogated by single nucleotide changes. Single mutations at positions +1 and +3 are the most consequential for Inr function, and the presence of a pyrimidine at position −1 is also important both in cells and in vitro (43, 57). Simultaneous mutation of these sites has more substantial effects. The gC Inr spans from −4 to +5 with one nucleotide mismatch (+4, Y to U) compared to the Inr consensus sequence (Fig. 1A). The LS mutant most affected (LS8) for transcription level, both in vitro and in virus infection, has mutations at −2, −1, and +1. The second-most-affected LS mutant (LS9) has mutations at positions +1 and +3 (Fig. 1B). Additionally, LS mutants 7 and 10 were also substantially affected in all the assays, and these have mutations at −2 and +1 and +3 and +5, respectively (Fig. 1B and 2). Therefore, the residues previously shown to be important for Inr function in other systems correlate with the residues important for gC expression in the virus and in vitro. Furthermore, the results also suggest that there are no other cis-acting sites affecting expression during infection in this region (−17 to +13).

Although mutations in the Inr sequence resulted in decreased basal transcription, ICP4-activated transcription was significantly more affected (Fig. 2 and 3). ICP4 still activated the LS8 template, albeit less efficiently than the templates with an intact Inr. The activation of the promoter by ICP4 in the absence of Inr function may reflect the ability of ICP4 to promote transcription preinitiation complexes (PICs) on the promoter by facilitating TFIID binding to the TATA box (27). The additional presence of a functional Inr and an activity of ICP4 may enhance PIC formation or trigger some postrecruitment step that results in full activation. While further studies are necessary to distinguish between these two possibilities, some reported data are relevant to this question.

TFIID binds to the TATA box via the TBP (for a review, see reference 33). However, TBP-associated factors (TAFs) in TFIID are involved in recognizing the Inr (6, 48, 58, 72, 87, 91, 93). In the Drosophila melanogaster system, a complex consisting of TBP-TAF150-TAF250 is sufficient for Inr activity. Evidence suggests that TAF250 recognizes the Inr and that dTAF150 (or the mammalian homologue CIF150) is required to support Inr activity (47, 49, 59, 88, ). TAFs also bridge transactivators and general transcription factors by protein-protein interactions, thus affecting initiation complex formation at the transcription start site, resulting in activation (57, 86). ICP4 requires TAFs for activation (28), and it interacts with TFIID in solution through TAF250 (8). The ability of ICP4 to interact with TAF250 is mediated by the carboxy-terminal 524 amino acids (8). Mutants that lack this region of ICP4 are significantly impaired in activation (80). The residual activation is a function of the amino-terminal 774 amino acids of ICP4 (16). This fragment of ICP4 is sufficient to promote PIC formation on the gC promoter (27).

Therefore, the amino terminus of ICP4 promotes PIC formation independent of an interaction with TAF250. The interaction with TAF250 as a consequence of some function in the carboxy-terminal region may allow TFIID to more efficiently function through the Inr. Thus, the interaction between ICP4 and TAF250 may enable the recognition of the Inr by TAF250 in a manner analogous to dTAF150 or CIF150, which may additionally involve a conformational change in the complex. Alternatively, the additive effects of additional multiple protein-protein and protein-DNA interactions as a function of ICP4, TAF250, the TATA box, and the Inr at the start site might drive the formation of an activated transcription complex. At present, we favor the former hypothesis. The latter hypothesis, while seemingly inconsistent with the PIC formation studies (27), warrants further testing considering the limitations of such studies.

Role of the Inr and leader region during viral infection.

The two important features of late genes that are in part functions of the activities of their promoters are their requirement for DNA synthesis and their abundant expression late after infection. The former question is a complex problem that has yet to be fully explained. The latter question is interesting in that the function of true late promoters is not apparently augmented by upstream activators, as is the case for IE and E genes. Therefore, the biological significance of the function of the Inr present in many late-gene promoters is an important question.

Several mechanisms have been suggested to explain late-gene expression during viral replication. These include differences in promoter strength and architecture, the presence of inhibitors, the conformational change of the genome after viral replication, and differences in polyadenylation sites (53, 62, 76, 82). It has been suggested that cis-acting elements in late genes play a role in temporal regulation. Furthermore, it was previously reported that the gC region from −35 to +124 in the tk locus results in late-expression kinetics and required DNA replication for expression to occur (35). In this study we found that the core gC promoter (−35 to +14) inserted in the tk locus did not result in true late-gene expression. However, the presence of a wt Inr in these constructs resulted in prolonged gene expression. Mutation of the Inr generally reduced expression and resulted in a gene expression profile more closely resembling that of early genes. Therefore, while the Inr does not mediate the requirement for DNA synthesis, it may contribute to true late-gene expression by allowing higher levels of expression late after infection.

When the core promoter was examined in the natural gC locus, expression was maintained late after infection. However, in contrast to the results obtained with the core promoter in the tk locus, expression was completely dependent on DNA synthesis. Moreover, the level of expression was very similar to that observed with the intact gC locus. Therefore, the region between +15 and +124 does not greatly affect the expression of the gC gene at the natural gC locus. This suggests that the context in which the promoter is situated and subsequently analyzed is very important. It may be that the requirement for DNA synthesis is not mediated by a definable set of cis-acting elements but is rather a function of both the promoter structure and the context of the promoter in the viral genome.

When the initiator element at the natural gC locus was mutated, expression kinetics were qualitatively unaltered relative to wt virus. However, expression was dramatically reduced. By comparison, the expression from the viruses containing the leader region deletion, or mutations in the potential DAS and DPE elements, were relatively unaffected compared to wt virus. In terms of sequence and location, DAS is highly similar to a downstream promoter element, DPE. DPE is a core promoter element mostly found in Drosophila TATA-less promoters (4, 7), and frequently functions in a combination with the Inr (3, 4). Having these additional core promoter elements (Inr and DPE) along with TATA box would be advantageous for expression of late genes. Our studies argue that while these elements may function in some late genes, they apparently contribute little to gC expression.

Our results suggest that true late-gene promoters greatly rely on the function of the Inr. Importantly, it is not simply the basal activity of the Inr that is important for high-level late-gene expression, but rather the ability of ICP4 to more efficiently activate transcription with a functional Inr. Late-gene expression and viral DNA replication occur in globular nuclear structures called replication compartments (13, 73), where specific viral proteins, including ICP4 (51), have been shown to localize. Thus, the compartmentalization of the viral genome and ICP4 and the ability of ICP4 to activate transcription more efficiently as a function of interactions with TFIID in the presence of an Inr, may collectively result in the high levels of late mRNA during infection. These high levels of expression are necessary to produce the large quantities of structural proteins required to assemble thousands of progeny virions per cell.