The Adenovirus L4 33-Kilodalton Protein Binds to Intragenic Sequences of the Major Late Promoter Required for Late Phase-Specific Stimulation of Transcription

Abstract

The adenovirus late IVa₂ protein is required for maximally efficient transcription from the viral major late (ML) promoter, and hence, the synthesis of the majority of viral late proteins. This protein is a sequence-specific DNA-binding protein that also promotes the assembly of progeny virus particles. Previous studies have established that a IVa₂ protein dimer (DEF-B) binds specifically to an intragenic ML promoter sequence necessary for late phase-specific stimulation of ML transcription. However, activation of transcription from the ML promoter correlates with binding of at least one additional infected-cell-specific protein, termed DEF-A, to the promoter. Using an assay for the DNA-binding activity of DEF-A, we identified the unknown protein by using conventional purification methods, purification of FLAG-tagged IVa₂-protein-containing complexes, and transient synthesis of viral late proteins. The results of these experiments established that the viral L4 33-kDa protein is the only component of DEF-A: the IVa₂ and L4 33-kDa proteins are necessary and sufficient for formation of all previously described complexes in the intragenic control region of the ML promoter. Furthermore, the L4 33-kDa protein binds to the promoter with the specificity characteristic of DEF-A and stimulates transcription from the ML promoter in transient-expression assays.

A characteristic feature of the infectious cycles of viruses with DNA genomes is DNA synthesis-dependent activation of transcription of viral late genes. In the case of human adenoviruses, such as adenovirus type 5 (Ad5), replication of the viral genome initiates a two-step transcriptional cascade. Transcription of the viral IVa₂ gene is first activated as a result of viral DNA synthesis-dependent titration of a cellular transcriptional repressor that binds to the IVa₂ promoter (10, 27, 36). Synthesis of the IVa₂ protein in infected cells then leads to maximally efficient transcription from the major late (ML) promoter, which controls expression of the coding sequences for all but one of the viral structural proteins (58). Entry into the late phase is accompanied by several changes in ML transcription. During the early phase, the ML and other early promoters are utilized with similar efficiencies (57), but ML transcription terminates at multiple sites within a large region in the middle of the transcription unit (1, 2, 28, 57). In contrast, termination occurs close to the right end of the viral genome during the late phase of infection (17). The difference in how much of the ML unit is transcribed, in conjunction with differences in the posttranscriptional processing of ML pre-mRNAs, results in production of only the L1 52/55-kDa protein early in infection but at least 15 ML mRNAs late in infection (15, 58). It has also been reported that the processivity of ML transcription beyond approximately position +1,000 increases late in infection (33). Finally, the efficiency of ML transcription increases by a factor of 20 to 30 once viral DNA synthesis has commenced (57).

The basal ML promoter comprises a typical TATA sequence, an initiator, GC-rich sequences near the initiator, and binding sites for the cellular proteins USF and CBF located upstream of the TATA sequence (8, 11, 42, 48, 50, 51, 55). Late phase-specific stimulation of ML transcription in vitro and in infected cells requires additional, intragenic sequences, termed DE1 (positions +86 to +96) and DE2 (positions +101 to +116) (29, 34, 41). These DE sequences are recognized by proteins present only in extracts of Ad5-infected cells harvested during the late phase of infection (29, 34, 44). Previous biochemical studies identified DEF-B, which binds to the DE2 sequence shown in Fig. 1, as a dimer of the IVa₂ protein (38, 60). This interaction was shown to stimulate ML transcription in transient-expression assays (60). The DE1 and DE2a sequences of the ML promoter (Fig. 1) are recognized by a second infected-cell-specific protein, termed DEF-A (30, 43). Initial efforts to purify DEF-A were not successful, although it was reported that the IVa₂ protein is also a component of this DNA-binding protein (38).

FIG. 1.

Electrophoretic mobility shift assay for IVa₂ and DEF-A. (A) The intragenic ML binding sites for the IVa₂ protein dimer (DEF-B) and DEF-A are illustrated schematically but to scale. (B) A ³²P-labeled double-stranded DNA containing these binding sites was incubated with no protein (0), whole-cell extracts prepared from mock-infected HeLa cells (U), or Ad5-infected cells harvested 24 h after infection (I). The protein-DNA complexes were examined as described in Materials and Methods. The infected-cell-specific a, b, and c complexes (44) are indicated on the right.

In addition to binding to the ML promoter, the IVa₂ protein recognizes repeated, redundant sequences termed A repeats (20, 21, 56) within the viral packaging signal (65) and is required for packaging of the viral genome during assembly (46, 66, 67). This protein is a component of infected-cell-specific complexes formed on A repeat sequences in vitro. Furthermore, the IVa₂ protein has been shown by chromatin immunoprecipitation to interact with the packaging sequence in infected cells (46, 49, 65).

The molecular mechanisms by which the IVa₂ protein stimulates ML transcription, and facilitates assembly are not understood. In addition to its sequence-specific DNA-binding activity, the IVa₂ protein interacts with the L1 52/55-kDa protein (22), which also plays an important role in genome packaging and assembly (23, 25). The L1 52/55-kDa protein has also been shown by chromatin immunoprecipitation to associate with the packaging signal in infected cells (46, 49). However, it does not bind to the ML promoter intragenic control region or to the packaging sequence, nor is it a component of IVa₂ protein-containing complexes formed on those viral sequences (49, 65). As the unknown identity of the protein(s) with which the IVa₂ protein cooperates to stimulate ML transcription represents a major impediment to a better understanding of this process, we set out to purify and identify this protein.

MATERIALS AND METHODS

Cells and virus.

HeLa, 293, and 293FT cells were maintained in monolayer cultures in Dulbecco's modified Eagle's medium (G1BCO-BRL) supplemented with 5% fetal bovine serum and 5% calf serum (Gemini). Ad5 was propagated in HeLa cells and titered by plaque assay on 293 cells (62).

Cell extracts and protein purification.

HeLa cells were infected with 20 PFU/cell Ad5, or mock infected, for 24 h unless otherwise indicated. Whole-cell extracts were prepared by the method of Manley et al. (40), except that the final dialysis was against 20 mM HEPES, pH 7.9, containing 5 mM MgCl₂; 50 mM KCl; 1 mM dithiothreitol (DTT); 1 μg/ml each of aprotinin, leupeptin, and pepstatin; 0.5 mM phenylmethylsulfonyl fluoride; and 20% (vol/vol) glycerol (DB). Nuclear extracts were prepared by lysing cells in 10 ml hypotonic lysis buffer (10 mM Tris-HCl, pH 7.5, containing 1.5 mM MgCl₂, 0.25 M sucrose, 1 mM DTT, and 0.5 mM phenylmethylsulfonyl fluoride) per 1 × 10⁷ to 2 × 10⁷ cells. Nuclei were pelleted by centrifugation at 2,850 × g for 10 min at 4°C, and the supernatant was discarded. The nuclei were extracted by incubation in 40 mM Tris-HCl, pH 7.5, containing 0.5 M NaCl; 0.2 mM EDTA; 1 mM DTT; 0.2 mM phenylmethylsulfonyl fluoride; 1 μg/ml each of aprotinin, leupeptin, and pepstatin; and 5% (vol/vol) glycerol. The nuclear extract was rotated at 4°C for 30 min and clarified by centrifugation for 20 min at 4,000 rpm. Nuclear proteins in the supernatant were precipitated with 65% saturated ammonium sulfate (pH ∼7.0) and pelleted by centrifugation at 30,500 × g for 20 min. The pellet was dissolved in DB and dialyzed against the same buffer.

About 3 mg extract protein were subjected to chromatography on a Superdex S-200 column (Pharmacia Biotech) in DB containing 0.1 M KCl and 10% (vol/vol) glycerol. Fractions were assayed by using the electrophoretic gel mobility shift assay described below. The peak fractions containing DEF-A activity were pooled, preincubated on ice for 1 h with 10 μg/ml poly(dI-dC) (Amersham Biosciences), and incubated with biotinylated DNA containing the DEF-A and DEF-B binding sites of the ML promoter (positions +78 to +126) conjugated to magnetic streptavidin beads (New England Biolabs) for 1 h at room temperature with rotation. The DNA affinity resin was washed twice with DB containing 0.1 M KCl for 20 min. The bound proteins were then eluted with DB containing 1.0 M NaCl and dialyzed against DB.

Mass spectrometry.

Protein bands were excised from sodium dodecyl sulfate (SDS)-polyacrylamide gels and analyzed by in-gel digestion with trypsin (Promega) and liquid chromatography-mass spectrometry/mass spectrometry. These analyses were performed by ProTech Inc. (Norristown, PA).

Electrophoretic gel mobility shift assay.

DNA oligonucleotides (Integrated DNA Technologies) were gel purified, 5′ end labeled with ³²P, and annealed as described previously (10, 31). One nanogram of labeled double-stranded DNA corresponding to positions +78 to +126 of the ML promoter was incubated with 2 to 5 μg of extract protein or chromatographic fractions, as indicated, in the presence of 420 ng (or less when appropriate) of fragmented poly(dI-dC) as a nonspecific competitor. Binding reactions (20 μl) contained 10 mM HEPES-KOH, pH 7.9, 50 mM KCl, 2 mM MgCl₂, 1 mM EDTA, and 2.5% (wt/vol) Ficoll. Competition assays also contained the molar-excess concentrations of unlabeled competitor DNAs indicated. After incubation of reaction mixtures for 15 min at room temperature, protein-DNA complexes were separated by electrophoresis in 4.5% polyacrylamide gels (80:1 acrylamide-bisacrylamide) cast and run in 45 mM Tris, 48 mM boric acid, pH 8.0, and 1 mM EDTA (1× Tris-borate-EDTA [TBE]) for 2 h at 260 V at 4°C. The gels were then dried and exposed to X-ray film.

Vectors for expression of viral coding sequences.

The IVa₂ and L4 33-kDa coding sequences were amplified by high-fidelity PCR using Accuprime Pfx (Invitrogen) with the Ad2 genome as a template and primers carrying restriction endonuclease sites. The sequences of the forward and reverse IVa₂ primers were GGATCGCGGCCGCGGATCCATGGAAACCAGAGGTAAGAAACGC and GCTTACCTCGAGAAGCTTTTACTATTTATCATCATCATCTTTATAATCTGATGCTTGTTTTGGGGTTTTGCGCGC. The latter included the sequence complementary to the FLAG coding sequences upstream of the complement of the IVa₂ stop codon. The sequences of the forward and reverse L4 33-kDa primers were GGATCGCGGCCGCGGATCCATGGCACCCAAAAAGAAGC and GCTTACATCGATTTACTAGTCCTTAAGAGTCAGCC. The L4 33-kDa sequence was also amplified with a forward primer carrying the coding sequence of the hemagglutinin (HA) tag. The amplified DNA fragments were cloned into the CβS expression vector (6), and the cloned viral genes were sequenced to ensure that they contained no PCR-generated mutations.

Transient expression and FLAG immunoprecipitation.

Ten micrograms of the expression vector DNAs described above were introduced into 293FT cells via the calcium phosphate method or Lipofectamine lipid reagent (Bio-Rad). Nuclear extracts were prepared as described above. For immunoprecipitation, nuclear extracts were incubated with anti-FLAG M2 agarose beads (Sigma) for 4 h at 4°C. The beads were then washed extensively with DB containing 0.1 M KCl on a minicolumn, and proteins were eluted with 2 bead volumes of DB containing 50 ng/ml of FLAG peptide (Sigma) for 12 h at 4°C. The L4 33-kDa protein carrying an N-terminal HA tag was purified in the same way, using anti HA-agarose beads (Sigma) and HA peptide (Sigma).

Isolation of stable 293FT cell lines producing IVa₂-FLAG protein.

The IVa₂-FLAG-containing vector described above and the hygromycin gene-containing vector pTK-hyg (CLONTECH) were introduced into 293FT cells via the calcium phosphate coprecipitation method. Hygromycin (100 μg/ml) was added 48 h after the introduction of DNA, and the cells were maintained in medium containing the drug for 2 weeks. Individual colonies that grew from single cells were then picked and expanded into cell lines. Cell lines that produced the IVa₂-FLAG protein at levels similar to that of the IVa₂ protein late during Ad5 infection were identified by immunoblotting of extracts, as described below.

Antibodies and immunoblotting.

The IVa₂-FLAG and HA-L4 33-kDa proteins were transiently synthesized in 293FT cells and purified by immunoaffinity chromatography as described above. The IVa₂ protein was further purified to apparent homogeneity by gel filtration on a Superdex S-200 column (Pharmacia Biotech) and concentrated using an Amicon filter (Millipore). The L4 33-kDa protein was subjected to SDS-polyacrylamide gel electrophoresis and eluted from gel slices by electroelution in 25 mM Tris, 0.192 M glycine, pH 8.3, containing 0.1% (wt/vol) SDS for 4.5 h. The eluate was dialyzed against phosphate-buffered saline (Gibco-BRL) containing 0.5% (vol/vol) Triton X-100. The purified proteins were injected into 8- to 12-week-old mice, with booster shots 2 and 5 weeks after the initial injection. The mice were bled 6 weeks after the initial injection. Each animal's response to antigen was tested by enzyme-linked immunosorbent assay with the purified proteins. The bleed served as a polyclonal antibody for the L4 33-kDa protein.

Unless otherwise indicated, proteins were separated by electrophoresis in 10% SDS-polyacrylamide gel electrophoresis gels and examined by immunoblotting, as described previously (19), using the M2 anti-FLAG monoclonal antibody (Sigma) or the antibodies against the L4 33-kDa and IVa₂ proteins.

Purification of cytoplasmic and nuclear RNAs.

293FT cells were harvested 44 h after introduction of the vector for expression of the L4 33-kDa protein coding sequences as described above. Nuclear and cytoplasmic fractions were separated by extraction with a buffer containing 0.65% (vol/vol) NP-40, and nuclear and cytoplasmic RNAs were purified as described previously (16). RNA pellets were dried and resuspended in 10 mM Tris-HCl, pH 7.5, containing 5 mM NaCl and 2 mM EDTA in water treated with diethylpyrocarbonate.

Analysis of RNA by RT-PCR.

For cDNA synthesis, 1 μg RNA was mixed in a 10-μl final volume with 15 pmol of an oligonucleotide complementary to a sequence just downstream of the L4 33-kDa protein mRNA splice site (see Fig. 5). The sequence of this reverse primer was GCTTACATCGATTTACTATCCGGTCGCC. The mixture was heated at 65°C for 10 min and cooled to 50°C. Reverse transcription (RT) reaction mixtures (20 μl) contained 1× reverse transcription buffer (Invitrogen), 0.1 mM DTT, 10 mM each deoxynucleoside triphosphate, 40 units RNasin (Promega), 1 μl reverse transcriptase (Invitrogen), and 10 μl of the RNA-primer or 10 μl H₂O. Reaction mixtures were incubated at 42°C for 1 h. PCR mixtures, which contained 2 μl cDNA, the reverse primer described above, and the L4 33-kDa forward primer described previously, were as described above. PCR products were analyzed by electrophoresis in 1% (wt/vol) agarose gels cast and run in 1× TBE and ethidium bromide staining.

FIG. 5.

The IVa₂ and L4 33-kDa proteins reconstitute DEF-A DNA-binding activity. (A, top) The structures of unspliced and spliced L4 33-kDa RNAs are depicted to scale, with the positions of the RT-PCR primers (not to scale) and the predicted products indicated below. As these two primers were also used for cloning L4 coding sequences, they added 37 bp to the predicted RT-PCR products. (Bottom) DNA products made in reactions that contained Ad2 DNA (Ad2) or cytoplasmic (C) or nuclear (N) RNA purified from 293FT cells containing the L4 33-kDa expression vector. Reactions that contained or lacked reverse transcriptase are indicated by + and −, respectively. A 100-bp DNA ladder (Invitrogen) was run in lane 1. The product of spliced L4 33-kDa RNA is indicated on the right. (B) Nuclear extracts were prepared from 293FT cells (lane 3) or from these cells 24 h after the introduction of vectors directing synthesis of the IVa₂ protein (lane 4), the L4 33-kDa protein (lane 5), or both vectors (lane 6) and assayed for binding to the intragenic ML promoter sequence. The reactions whose products are shown in lanes 2 and 7 contained an Ad5-infected cell nuclear extract (I) and both the IVa₂-containing and the L4 33-kDa-protein-containing extracts, respectively. (C) Reactions contained the extract isolated from cells producing both the IVa₂ and the L4 33-kDa proteins and no competitor (lane 1), a 30-fold molar excess of DE m1 competitor that retained the wild-type DE2 sequence (lane 2), or a competitor (DE m12ab) in which all three DE sequences were mutated (lane 3). (D) Binding reactions contained nuclear-extract proteins from 293FT cells synthesizing the viral proteins indicated at the top (lanes 3 to 6), no protein (lane 1), or Ad5-infected nuclear extract (I; lane 2) and ³²P-labeled DEm1 DNA. This probe was identical to the wild-type probe, except for the presence of DE1 (Fig. 1A) substitutions that block binding of DEF-A (37).

Transient-expression assays.

A reporter gene comprising the enhanced green fluorescent protein (EGFP) coding sequence under the control of the ML promoter (positions −150 to +220) (10 μg) was introduced into 293 cells alone or with 10 μg DNA of the L4 33-kDa or IVa₂ protein expression vectors described above. The cells were harvested 24 h later. Total RNA was purified by using Trizol (Invitrogen) according to the manufacturer's protocol and treated with 40 units RNase-free DNase I (Roche Diagnostics) for 1 h at 37°C. The RNA was then reextracted with Trizol and ethanol precipitated. The concentrations of primary ML transcripts were determined by using RT-PCR with primers that spanned the ML sequence that was included in the reporter gene. Total RNA (3.0 μg) was amplified and labeled by using One-Step RT-PCR (Roche Diagnostics; 16 cycles) in reaction mixtures containing 3 μCi [α-³²P]dATP (3,000 Ci/mol; Perkin-Elmer). The labeled DNAs were analyzed by electrophoresis in 6% polyacrylamide gels cast and run in 0.25× TBE and autoradiography.

RESULTS

We initially attempted to identify the unknown protein component(s) of DEF-A by using conventional purification methods and following the DNA-binding activity of DEF-A using an electrophoretic mobility shift assay. Nuclear or whole-cell extracts, or fractions recovered during chromatography, were incubated with ³³P-labeled double-stranded DNA corresponding to the DE region of the viral ML promoter (positions +78 to +126). As illustrated in Fig. 1B, proteins present in whole-cell extracts prepared from HeLa cells harvested 24 h after Ad5 infection formed three complexes that were not detected when reaction mixtures contained proteins from uninfected cells. The complexes migrated with the relative mobilities of the previously described a, b, and c complexes, which contain DEF-A, DEF-B, and both DEF-A and DEF-B, respectively (44). Competition experiments with unlabeled DNAs carrying mutations in the individual binding sites confirmed the identities of these complexes (see Fig. 5B and data not shown).

Most combinations of chromatographic procedures that we tested resulted in loss of DEF-A activity, as also reported by Lutz and Kedinger (38). However, gel filtration followed by sequence-specific DNA affinity chromatography (Fig. 2A) allowed substantial purification of active DEF-A. When nuclear extracts of Ad5-infected 293FT cells were fractionated by chromatography on a Superdex S-200 column, the infected-cell-specific complexes were recovered in two major peaks (Fig. 2B). The first peak (peak 1 in Fig. 2B) contained DEF-A, as judged by the formation of complexes a and c, but at lower concentrations than the IVa₂ protein dimer (DEF-B), which was present in complex b. In contrast, the second peak was more enriched for DEF-A (compare, for example, fractions 40 and 64 in Fig. 2B). Furthermore, most of the cellular proteins that could bind to the internal ML promoter sequence eluted before peak 2 of DEF-A activity (Fig. 2B), as did most proteins (data not shown). We therefore pooled the later-eluting DEF-A-containing fractions and subjected them to DNA affinity chromatography using a biotinylated derivative of the ML internal control region bound to streptavidin beads. Both the IVa₂ protein dimer (complex b) and DEF-A (complexes a and c) were recovered when the DNA-bound proteins were eluted with a buffer containing 1.0 M NaCl (Fig. 2C). Examination of the proteins present in the various fractions by SDS-polyacrylamide gel electrophoresis and silver staining established that a protein of some 33-kDa apparent molecular mass was also enriched in the 1 M NaCl eluate (Fig. 2D). Purification of extracts of uninfected 293FT cells stably synthesizing the IVa₂ protein by the same two steps yielded a 1.0 M NaCl eluate of the DNA affinity resin that contained the IVa₂ protein but not the 33-kDa protein (data not shown). Immunoblotting with polyclonal antibodies raised against a viral protein transiently synthesized from the L4 33-kDa protein coding sequence identified the 33-kDa protein enriched by DNA affinity chromatography as a product of this viral L4 gene (Fig. 2E). Although the flowthrough fraction contained no DEF-A DNA-binding activity (Fig. 2C), the L4 protein was also detected in this fraction (Fig. 2E). However, as discussed below, different forms of the L4 protein were recovered in the flowthrough and 1.0 M NaCl fractions.

FIG. 2.

Purification of DEF-A. (A) Summary of the purification scheme. (B) Electrophoretic mobility shift assay of fractions recovered from the Superdex S-200 column. The positions of the infected-cell-specific complexes are indicated on the right. The positions of the two peaks of DEF-A activity described in the text are indicated below the fraction numbers. Fractions 62 to 69 were pooled and purified by affinity chromatography on DNA containing the ML intragenic promoter sequences. (C) Electrophoretic mobility shift assay of the pooled S-200 fractions (S) and fractions recovered during DNA affinity chromatography (U, unbound proteins; W, proteins recovered in 0.1 M wash buffer; E, proteins recovered in 1.0 M NaCl). (D) Silver staining of the DNA affinity chromatography fractions, indicated as in panel C, following separation by electrophoresis in a 10% SDS-polyacrylamide gel. Prestained molecular mass markers, as listed at the left, were loaded in the lane marked M. The positions of the IVa₂ and enriched 33-kDa proteins are indicated on the right. (E) The same DNA affinity chromatography fractions were examined by immunoblotting using monoclonal and polyclonal antibodies specific for the IVa₂ and L4 33-kDa proteins, respectively.

It has been reported that the L4 33-kDa protein migrates more slowly than predicted from its molecular mass in SDS-polyacrylamide gels and that an additional protein, termed L4 22 kDa, is encoded by the same L4 gene (32, 45). The 33- and 22-kDa proteins, which are synthesized from spliced and unspliced mRNAs, respectively (45), share an N-terminal sequence of 105 amino acids but possess unique C-terminal regions. To determine which of these L4 proteins copurified with DEF-A DNA-binding activity, the protein of 33-kDa apparent molecular mass shown in Fig. 2D was subjected to mass spectrometry (Table 1). In addition to a peptide common to the shared N-terminal segments of the 33-kDa and 22-kDa proteins, only peptides present in the unique C-terminal sequence of the 33-kDa protein were detected. We therefore conclude that the L4 33-kDa protein copurified from Ad5-infected extracts with DEF-A DNA-binding activity and that the polyclonal antibodies used in these experiments (Fig. 2E) specifically recognize this L4 protein.

TABLE 1.

L4 33-kDa peptides identified by mass spectrometry^a

Peptide	Sequence	L4 protein (kDa)
1	TLEDAELFSK	33
	IFPTLYAIFQQSR	33
2	SATGSSMATTSAPQAPPALPVR	33 and 22
	TLEDAEALFSK	33
	IFPTLYAIFQQSR	33

^aThe results of two analyses of the 33-kDa protein (Fig. 2D) are shown.

The L4 33-kDa protein binds to the IVa₂ protein.

The results described above identified the viral L4 33-kDa protein as a likely DEF-A candidate. To obtain additional evidence for this conclusion, we wished to purify IVa₂ and associated proteins by a different method. We therefore isolated lines of 293FT cells that synthesized a derivative of the IVa₂ protein carrying a C-terminal FLAG tag (IVa₂-FLAG) from integrated coding sequences, as described in Materials and Methods. The concentrations of IVa₂ protein produced in the various clonal cell lines obtained were compared to that present during the late phase of Ad5 infection by immunoblotting (data not shown). Subsequent experiments were performed in a line in which the IVa₂ protein accumulated to a steady-state concentration similar to that attained by 24 h after Ad5 infection. As expected, extracts of the uninfected cells contained only DEF-B, but Ad5 infection led to the formation of complex c (Fig. 3A, lane NE). We therefore purified IVa₂-FLAG protein from Ad5-infected cells by immunoprecipitation with an anti-FLAG antibody. The bound proteins were eluted by competition with FLAG peptide, as described in Materials and Methods, and assayed as described above. Both DEF-A and DEF-B were immunoprecipitated (Fig. 3A, lane IP), and the L4 33-kDa protein again coeluted with DEF-A activity (Fig. 3B), indicating that it interacts with the IVa₂ protein.

FIG. 3.

Copurification of DEF-A and the L4 33-kDa protein by binding to a FLAG-tagged IVa₂ protein. Nuclear extracts prepared from mock- or Ad5-infected 293FT cells stably synthesizing the IVa₂-FLAG protein were subjected to anti-FLAG immunoaffinity chromatography as described in Materials and Methods. (A) The Ad5-infected cell extract (NE), wash (W), and immunoprecipitated proteins (IP) were assayed for proteins that bound to the ML DE sequence. The positions of the b and c complexes are indicated on the right. (B) The nuclear extract and immunoprecipitated proteins (IP) from the experiment shown in panel A (Ad5) or from mock-infected cells (Mock) were examined by immunoblotting them with polyclonal antibodies against the L4 33-kDa protein.

As the L4 33-kDa protein copurified with DEF-A DNA-binding activity through two very different procedures, we next examined the effects of antibodies against this protein on binding of infected-cell-specific proteins to the ML intragenic promoter sequence. Whole-cell extracts prepared from Ad5-infected cells were incubated with the mouse polyclonal antibody against the L4 33-kDa protein described in the previous section, or with monoclonal antibodies that recognize the IVa₂ protein or viral protein V, for 15 min prior to addition to DNA-binding reaction mixtures. The control anti-protein V antibody had no effect on the formation of either the b or c complex detected under the conditions used (Fig. 4, lanes 1 and 4). In contrast, the anti-IVa₂ protein antibody impaired the formation of both complexes (Fig. 4, lane 3), as expected, whereas the anti-L4 33-kDa protein antibody specifically blocked the formation of complex c but had no effect on binding of the IVa₂ protein (complex b) (Fig. 4, lane 2). We therefore conclude that, like DEF-A, the L4 33-kDa protein is an essential component of complex c.

FIG. 4.

Inhibition of binding of DEF-A to the ML promoter by antibodies against the L4 33-kDa protein. A whole-cell extract prepared from HeLa cells 24 h after Ad5 infection was preincubated on ice for 15 min with no (−) antibody or with (+) the antibody indicated at the top. Binding of proteins to the ML DE sequence was then assayed. The positions of the b and c complexes are indicated on the right.

The L4 33-kDa protein is necessary and sufficient to reconstitute DEF-A.

The results described above are consistent with the conclusion that the L4 33-kDa protein is a component of DEF-A. However, as noted in the introduction, other viral proteins have been reported to interact with the IVa₂ protein, raising the possibility that DEF-A might contain additional proteins. To address this issue, we examined the DNA-binding activities present in cells synthesizing only the IVa₂ or the L4 33-kDa protein or both proteins. Vectors carrying the coding sequences for the IVa₂-FLAG or the L4 33-kDa protein under the control of the human cytomegalovirus immediate-early protein were introduced into 293FT cells. The L4 vector contained the complete 33-kDa protein coding sequence and could, therefore, direct synthesis of both the spliced mRNA for this protein and the unspliced mRNA specifying the 22-kDa protein (45). We therefore used RT-PCR to examine the nuclear and cytoplasmic L4 RNAs made in cells containing this expression, using the primers illustrated in Fig. 5A. The RT-PCR product predicted for the complete L4 coding sequence spanned by these primers was readily detected when genomic DNA was used as a template in PCRs (Fig. 5A, lane 2). This product was also observed when RT reaction mixtures contained nuclear RNA, but not when nuclear RNA was examined directly by PCR (Fig. 5A, lanes 5 and 6). These results established that, as expected, unspliced L4 mRNA was present in the nuclei of cells containing the L4 expression vector. In contrast, only the RT-PCR product corresponding to spliced 33-kDa mRNA was detected when cytoplasmic mRNA was used as a template, with no trace of unspliced L4 RNA evident (Fig. 5A, lane 4). We therefore conclude that under the conditions used in these experiments, only the spliced L4 mRNA, and hence, only the L4 33-kDa protein, was produced in 293FT cells as described above.

Cells were harvested 24 to 36 h after the introduction of DNA, and nuclear extracts were prepared and assayed for proteins that bound to the ML DE promoter sequence. When the IVa₂-FLAG protein alone was made, only complex b could be detected (Fig. 5B, lane 4), as observed in cell lines stably synthesizing this protein. When both proteins were synthesized in 293FT cells, or when extracts prepared from cells making the IVa₂-FLAG or the L4 33-kDa protein were mixed, the DEF-A-containing complexes a and c were also formed (Fig. 5B, lanes 6 and 7). These results demonstrate that the viral L4 33-kDa protein and the IVa₂ protein are the only components required to form the a, b, and c complexes in the DE region of the ML promoter. Unexpectedly, however, the L4 33-kDa protein itself bound to this sequence to form complex a (Fig. 5B, compare lanes 2 and 5).

To confirm the identities of the complexes detected in these experiments, we performed competition experiments. Unlabeled DNA that retained the DE2 binding sites for DEF-A and DEF-B (Fig. 1) effectively blocked the formation of complexes a, b, and c, whereas unlabeled DNA carrying mutations in all three binding sites (Fig. 1) did not (Fig. 5C). In addition, we examined binding of the proteins to labeled ML DNA carrying substitutions that prevented binding of DEF-A to the DE1 sequence (44). The IVa₂ protein bound to this probe (DEm1), just as it did to the wild-type DNA (Fig. 5D, lane 4). However, the DE1 mutations virtually abolished binding of the L4 33-kDa protein to form complex a (Fig. 5D, lane 5) but did not prevent the formation of complex c in reactions that contained both viral proteins (Fig. 5D, lane 6). These results not only confirm that the L4 33-kDa protein binds specifically to ML promoter sequences, but also establish that it exhibits exactly the same DNA-binding properties as DEF-A (44).

Coproduction of the IVa₂ protein and a derivative of the L4 33-kDa protein carrying an N-terminal HA tag also allowed the formation of the DEF-A-containing complex c (data not shown). However, the HA-L4 33-kDa protein alone was incapable of binding to DNA to form complex a (data not shown). Consequently, complex c must have formed as a result of interaction of the HA-L4 33-kDa protein with DNA-bound IVa₂ protein.

A hypophosphorylated form of the L4 33-kDa protein binds to DNA.

It is well established that the L4 33-kDa protein synthesized in Ad5-infected cells is phosphorylated (3, 18, 53). To investigate whether this modification influences the ability of the protein to bind to DNA, we initially examined the kinetics of synthesis of the L4 33-kDa protein and of DEF-A DNA-binding activity in Ad5-infected HeLa cells. Both the L4 protein, which was detected by immunoblotting, and DEF-A activity were first detected 14 h after infection and increased significantly in concentration by 24 h (Fig. 6A). At 14 h after infection, a single species of the L4 33-kDa protein (Fig. 6A) was observed (even upon overexposure of immunoblots), while a second, more slowly migrating form was evident by 24 h (Fig. 6A and B). The synthesis of the more rapidly migrating form of the L4 33-kDa protein and DEF-A by 14 h after infection indicated that this form of the protein is competent to bind DNA. Consistent with this conclusion, one of two forms of the L4 33-kDa protein produced by transient synthesis in uninfected cells comigrated with the more rapidly migrating form of the protein made in infected cells (Fig. 6B and C). Furthermore, the L4 33-kDa protein that bound to ML DE DNA during sequence-specific DNA affinity chromatography migrated more rapidly than the L4 protein that did not bind (Fig. 2E).

FIG. 6.

DEF-A DNA-binding activity is associated with a hypophosphorylated form of the L4 33-kDa protein. Whole-cell extracts were prepared from HeLa cells infected with Ad5 for the times indicated or mock infected (M) and examined for proteins that bind to the ML DE sequence (A) or by immunoblotting for the L4 33-kDa protein (B). The positions of the a, b, and c complexes are indicated on the right of panel A. Extracts prepared from 293FT cells transiently synthesizing the L4 33-kDa protein were loaded onto the lanes marked L4 in panel B before (−) or after (+) treatment with 400 units phosphatase (PPase) for 1 h at 30°C, followed by incubation for 1 h at 65°C. (C) Proteins present in the 24-h-infected cell extract (I) and that prepared from L4 33-kDa protein-producing 293FT cells (L4) shown in panel A were separated by electrophoresis in 15% SDS-polyacrylamide gel electrophoresis and examined by immunoblotting them with polyclonal antibodies against the L4 33-kDa protein. (D) The infected cell extract (I) was examined as described above before (−) and after (+) treatment with phosphatase. p.i., postinfection.

To determine whether the two forms of the L4 33-kDa protein detected in infected cells were differentially phosphorylated, we examined the effects of treatment with lambda phosphatase. As illustrated in Fig. 6D, exposure to this enzyme converted the more slowly migrating to the more rapidly migrating species, indicating that the former was more heavily phosphorylated. However, the migration of the transiently synthesized L4 33-kDa proteins was not altered by treatment with enzyme, and the more rapidly migrating species produced under these conditions was not observed in Ad5-infected cells (Fig. 6B). We therefore conclude that a hypophosphorylated form of the L4 33-kDa proteins binds to the ML DE promoter sequences.

The L4 33-kDa protein stimulates transcription from the ML promoter.

To determine whether the L4 33-kDa protein can stimulate transcription from the ML promoter, we examined its effects on the expression of a reporter gene under the control of this viral promoter. A plasmid containing the EGFP coding sequence linked to a DE-containing ML promoter (Fig. 7A) was introduced into 293 cells alone or with the expression vectors for the L4 33-kDa and/or IVa₂ protein described previously. We then examined synthesis of RNA from the reporter gene, using RT-PCR with a reverse primer complementary to ML sequences downstream of the DE region. When present alone in cells, the ML promoter directed quite efficient synthesis of the reporter RNA (Fig. 7B, lane 2). Nevertheless, the L4 33-kDa protein induced a substantial increase in the reporter RNA concentration, either alone or when the IVa₂ protein was also made (Fig. 7B, lanes 4 and 5). The effect of the latter protein alone was less pronounced than that of the L4 33-kDa protein (Fig. 7B, compare lanes 3 and 4). As the reporter gene contained a heterologous poly(A) addition site and 3′ untranslated region and lacked a complete ML intron, we conclude that the L4 33-kDa protein stimulates transcription from the ML promoter.

FIG. 7.

The L4 33-kDa protein stimulates transcription from the ML promoter. (A) The organization of the reporter gene is shown schematically, with the ML promoter and EGFP coding sequences represented by the solid line and open box, respectively. The positions of the site of initiation of ML transcription (gray arrow), the DE sequence (gray box), and the primer for reverse transcription (dashed arrow) are indicated. (B) Ten micrograms of the plasmid containing the ML-GFP reporter gene were introduced into 293FT alone (−) or with (+) 10 μg of IVa₂ or L4 33-kDa expression plasmid, as indicated at the top. Total RNA was purified 27 h later and analyzed by RT-PCR with primers specific for the ML sequence of the reporter gene (ML) or human β-actin as a control for cell number and RNA recovery.

The L4 33-kDa protein associates with the viral packaging signal only in the presence of the IVa₂ protein.

As discussed in the introduction, the IVa₂ protein also binds specifically to repeated sequences of the viral packaging signal, and both it and the L4 33-kDa protein are required for assembly of virus particles. These properties suggested that the L4 33-kDa protein might also associate with the viral packaging signal. We therefore examined the complexes formed on the AI-AII repeat of the packaging signal by proteins present in extracts of cells synthesizing the IVa₂ and/or L4 33-kDa protein, as described above. In agreement with a previous report (65), we observed two infected-cell-specific complexes when the AI-AII repeat was used as a probe (Fig. 8, compare lanes 2 and 3). Both of these complexes, which have been designated x and y, have been reported to contain the IVa₂ protein (65). When reaction mixtures contained extracts of cells producing only the IVa₂ protein, complex y was found (Fig. 8, lane 4). In contrast, neither infected-cell-specific complex was detected using L4 33-kDa protein-containing extracts (Fig. 8, lane 5). However, production of the two viral proteins in 293FT cells resulted in the formation of complex x (Fig. 8, lane 6). The inability of the L4 33-kDa protein alone to bind to AI-AII packaging-repeat DNA is not a trivial consequence of too low a protein concentration, as the same L4 33-kDa protein-containing extracts were used in the experiments whose results are shown in Fig. 5. Unlabeled DNA containing wild-type DE2 sequences of the ML promoter effectively blocked the formation of complexes x and y, whereas the mutated ML promoter sequence did not (Fig. 8, lanes 7 and 8). We therefore conclude that the L4 33-kDa protein can interact with the viral packaging sequence repeat, but only when the IVa₂ protein is bound to this DNA.

FIG. 8.

The L4 33-kDa protein interacts with a packaging-signal repeat sequence only in the presence of the IVa₂ protein. Binding reactions contained no protein (0), nuclear extract prepared from HeLa cells infected with Ad5 for 24 h (I), or extracts isolated from 293FT cells synthesizing the viral proteins indicated at the top and 1 ng of the AI-AII packaging-repeat sequence labeled as described in Materials and Methods. A 30-fold molar excess of the unlabeled DE m1 and DE m12ab competitor DNAs described in the legend to Fig. 5 were included in the reactions whose products are shown in lanes 7 and 8, respectively.

DISCUSSION

Although DEF-A was described in 1992 as an infected-cell-specific protein that binds cooperatively to the ML DE sequence with the IVa₂ protein (DEF-B) (44), its molecular identity has remained elusive. The data reported here establish that DEF-A is the viral L4 33-kDa protein: DEF-A DNA-binding activity and the L4 33-kDa protein copurified through two entirely different protocols (Fig. 2 and 3 and Table 1), antibodies that recognize differentially phosphorylated forms of the L4 33-kDa protein (Table 1 and Fig. 2E and 6) blocked the formation of DEF-A-containing complexes by Ad5-infected cell proteins (Fig. 4), and synthesis of the L4 33-kDa and IVa₂ proteins in uninfected cells reconstituted the full set of complexes formed by infected cell proteins on the ML DE sequence (Fig. 5). Furthermore, when synthesized in the absence of any other viral proteins, this L4 protein formed complex a by binding to the ML DE1 sequence and, in the presence of the IVa₂ protein, bound to the DE2a sequence in complex c (Fig. 5). Thus, the L4 33-kDa protein exactly reproduced the previously described DNA-binding specificity of DEF-A (44).

The latter finding was unexpected, as Lutz and Kedinger (38) reported that the IVa₂ protein is a component of DEF-A. This conclusion was based, in part, on the reappearance of IVa₂ DNA-binding activity (DEF-B) when Ad5-infected cell fractions that contained only DEF-A were subjected to additional chromatographic procedures, a phenomenon interpreted as dissociation of the IVa₂ protein from a relatively unstable DEF-A heteromer. Subsequently, the IVa₂ protein has been reported to interact with other viral proteins in infected cells, including the L1 52/55-kDa protein (22), the major core protein VII and its precursor (64), and in the case of the porcine adenovirus type 3 protein, protein VIII (59). Initial copurification of any of these IVa₂ protein-containing complexes with DEF-A, followed by dissociation during subsequent chromatographic steps, would account for the reappearance of DEF-B during purification of DEF-A (38). It was also reported that polyclonal and monoclonal antibodies raised against an N-terminal fragment or the full-length IVa₂ protein, respectively, blocked or impaired the DNA-binding activity of DEF-A (38). As it is now clear that DEF-A does not contain the IVa₂ protein, such inhibition must have been nonspecific or the result of cross-reaction of the antibodies with the L4 protein.

Transcription of the IVa₂ gene is activated upon viral DNA synthesis-dependent titration of a cellular repressor (10, 27, 36), making the IVa₂ protein available soon after the onset of the late phase. In contrast, the proposition that a late protein encoded within the ML transcription unit participates in ML transcription appears paradoxical: the L4 33-kDa protein cannot be made until the late phase-specific switches to transcription of the complete ML transcription unit and utilization of all polyadenylation sites (see the introduction) have taken place. Although the mechanisms responsible for these changes in expression of the ML transcription unit are not understood, various observations indicate that the L4 proteins are made in infected cells before other ML gene products. When the synthesis of viral proteins was examined in Ad2-infected HeLa cells by pulse-labeling, the L4 33-kDa protein was seen to attain its maximal concentration near the beginning of the late phase, while other ML proteins accumulated later (4). In addition, analysis of the requirements for viral DNA replication and protein synthesis during the transition into the late phase identified a period in which the L4 mRNAs (as well as the L1 52/55-kDa mRNA that is the sole product of ML transcription early in infection [58]) were produced, but the other ML mRNAs were not (33). This pattern required viral DNA synthesis but no, or very little, synthesis of late proteins (33). One interpretation of these observations is that synthesis of progeny viral DNA molecules, which are not associated with viral core proteins, in contrast to the templates for transcription of viral early genes (9, 24, 63), is sufficient for transcription to the end of the ML transcription unit and that utilization of the L4 polyadenylation site is initially favored. Additional experiments will clearly be necessary to elucidate the mechanisms by which the L4 33-kDa mRNA and protein are made prior to the appearance of the full panoply of ML mRNAs in infected cells.

It was also reported that production of the full set of the L1 to L5 mRNAs in infected cells depended on protein synthesis during the late phase of infection (33). More recently, Farley and colleagues have demonstrated that the L4 33-kDa protein is both necessary and sufficient for synthesis of all ML mRNAs and proteins (12). This conclusion was based on analysis of the viral gene products made from ML transcription units of different lengths expressed in uninfected cells from a tetracycline repressor-regulated promoter. When all five regions of the transcription units were included, all L1 to L5 ML mRNAs and proteins were synthesized efficiently (12). However, truncation of the ML transcription unit just downstream of the L3 polyadenylation site resulted in the early pattern of ML expression, (12). Synthesis of the L4 33-kDa protein in cells containing this truncated ML gene was sufficient to induce the switch to the late pattern of ML mRNA synthesis (12). Furthermore, examination of the accumulation of processed L1 to L3 mRNAs in the cytoplasm and of L3 pre-mRNA in the nucleus indicated that the 33-kDa protein acted at a posttranscriptional step to increase mRNA production (12).

This posttranscriptional function of the L4 protein appears to be distinct from its ability to stimulate expression of the ML-EGFP reporter gene illustrated in Fig. 7: we observed the production of increased quantities of unprocessed transcripts from the ML promoter in the presence of the L4 33-kDa protein and both this and the IVa₂ protein. Furthermore, although our reporter gene included the endogenous ML promoter and part of the first intron of this viral gene, it lacked all other sites of posttranscriptional processing of ML pre-mRNAs, whereas those used by Farley et al. (12) lacked the upstream sequences of the ML promoter that cooperate with the DE sequence to stimulate ML transcription (43, 44). We therefore propose that the L4 33-kDa protein contributes to optimal expression of the ML transcription unit by both transcriptional and posttranscriptional mechanisms. In this respect, it appears to resemble cellular proteins that participate in both transcription of protein-coding genes by RNA polymerase II and the processing or export of processed transcripts, such as the THO/TREX complex (5, 35, 39, 52, 61).

The results of previous genetic experiments have implicated the proteins that bind to the ML DE sequence in efficient transcription from the ML promoter: mutations in DE sequences that block binding of infected-cell-specific proteins in vitro were found to be lethal when combined with mutations in the upstream binding site for USF in the ML promoter (47). However, as the binding sites for both the IVa₂ and the L4 33-kDa proteins were mutated (47), the properties of this mutant do not provide any information about the contributions of the individual viral proteins to ML transcription. Termination codons have been introduced either early in the L4 33-kDa protein coding sequence (13, 32) or 47 amino acids before the natural stop codon in the viral DNA genome (14). Although the effects of these mutations on virus viability varied, all induced phenotypes consistent with defects in the assembly of virus particles: progeny viral genomes and late proteins were synthesized, but production of virus particles was impaired or eliminated (13, 14, 32). However, these studies did not address the role of the 33-kDa protein in ML transcription. In the first place, truncated L4 33-kDa proteins that potentially retain some function could be produced from each of the substituted coding sequences, as observed directly in one case (32). Furthermore, in no case were the effects of the mutations on ML transcription examined directly. It is also possible that any contribution of the L4 33-kDa protein to efficient ML transcription during the late phase of infection can be detected only when the ML promoter carries substitutions in the upstream binding site for USF, as in the experiments of Pardo-Mateos and Young (47) described above, or when neither the IVa2 nor the L4 33-kDa proteins can be synthesized. An important priority, therefore, is to construct mutant viral genomes that allow the role of the L4 33-kDa protein in ML transcription to be assessed unambiguously.

While this article was in preparation, Ostapchuk and colleagues reported that, in contrast to our data (Fig. 5), the L4 33-kDa protein transiently synthesized in HeLa cells that stably produce the IVa₂ protein does not bind to the ML DE sequence (45). This difference may be the result of differences in the concentrations of the L4 33-kDa proteins attained; although the human cytomegalovirus immediate-early promoter was used to direct expression of L4 33-kDa genes in both sets of experiments, our construct included the intron present in this coding sequence, whereas that of Ostapchuk et al. (45) did not. It is well established that inclusion of introns in genes to be expressed from plasmid vectors in mammalian cells increases the efficiency of processing and export of mRNAs (7, 26, 37, 54). In this context, it may be significant that Ostapchuk et al. (45) did not detect the formation of complex b by the IVa₂ protein alone, suggesting that the concentration of the protein was low, a property that could also contribute to the failure to detect cooperative binding of the 33-kDa and IVa₂ proteins to the ML promoter sequence. It is also possible that the L4 33-kDa proteins were differentially phosphorylated as a result of synthesis in different cell types (293FT or HeLa cells) and/or differences in their intracellular concentrations. As discussed above, it is a hypophosphorylated form of the L4 33-kDa protein that binds the ML DE sequences.

It was also reported that an L4 22-kDa protein, which shares its N-terminal 105 amino acids with the 33-kDa protein, did bind to the DE sequence, as well as to a packaging-signal repeat, in the presence of the IVa₂ protein (45). If this protein binds to ML DE DNA with the same specificity as the 33-kDa protein, which is not yet clear, this observation implies that a DNA-binding domain is present within the N-terminal segment common to the two proteins. The 22-kDa protein has been ascribed a role in assembly on the basis of both its binding to the packaging sequence and the phenotypes induced by the introduction of a termination codon within the coding sequence that is expressed only in this protein (45). Despite its ability to bind to the ML DE sequence in the presence of the IVa₂ protein, we do not believe that the L4 22-kDa protein is likely to contribute to regulation of ML transcription in infected cells: the polyclonal antibodies that prevented binding of DEF-A to the DE1 and DE2a sequences required for late phase-specific stimulation of ML transcription (Fig. 4) detected only differentially phosphorylated forms of the L4 33-kDa protein in infected cells (Table 1 and Fig. 6). It is therefore possible that, although the L4 33-kDa and 22-kDa proteins share sequences, they differ in their DNA-binding capabilities and abilities to stimulate ML transcription. This hypothesis is consistent with our observation that the L4 33-kDa protein alone cannot bind to packaging-repeat sequences (Fig. 8) but does bind to the DE sequence (Fig. 5). Experiments are in progress to compare directly the DNA-binding and other properties of the L4 33-kDa and 22-kDa proteins.