Requirement of the Adenovirus IVa2 Protein for Virus Assembly

Abstract

The adenovirus L1 52/55-kDa protein is required for viral DNA packaging and interacts with the viral IVa2 protein, which binds to the viral packaging sequence. Previous reports suggest that the IVa2 protein plays a role in viral DNA packaging and that this function of the IVa2 protein is serotype specific. To further examine the function of the IVa2 protein in viral DNA packaging, a mutant virus that does not express the IVa2 protein was constructed by introducing two stop codons at the beginning of the IVa2 open reading frame in a full-length bacterial clone of adenovirus type 5. The mutant virus, pm8002, was defective for growth in 293 cells, although it replicated its DNA and produced early and late viral proteins. Electron microscopic and gradient analyses revealed that the mutant virus did not assemble any viral particles in 293 cells. In 293-IVa2 cells, which express the IVa2 protein, infectious viruses were produced, although the titer of the mutant virus was lower than that of the wild-type virus, indicating that these cells may not fully complement the mutation. The mutant viral particles produced in 293-IVa2 cells were heterogeneous in size and shape, less stable, and did not traffic efficiently to the nucleus. Marker rescue experiments with a wild-type IVa2 DNA fragment confirmed that the only mutations present in pm8002 were in the IVa2 gene. The results indicate that the IVa2 protein is required for adenovirus assembly and suggest that virus particles may be assembled around the DNA rather than DNA being packaged into preformed capsids.

The adenovirus capsid is an icosahedron containing three major proteins, hexon (720 molecules/virion), penton base (60 molecules/virion), and fiber (36 molecules/virion) (55, 61, 62), along with a series of minor components. Inside the capsid is the viral genome with associated core proteins (47). During assembly, hexon proteins trimerize to form hexon capsomers, which are the major structural units of the capsid (45). These capsomers then come together with the other capsid proteins and scaffolding proteins to form the capsid. The mechanisms of how adenovirus particles assemble and viral DNA is packaged are not yet fully understood. Empty capsids, which contain no viral DNA or core proteins, are produced in parallel to mature virions in infected cells (57). Intermediate particles with buoyant densities between those of empty capsids and mature virions have also been isolated from infected cells. These intermediates can be divided into light (ρ = 1.315 g/cm³) particles containing small DNA fragments and heavy (ρ = 1.37 g/cm³) particles containing full-length DNA (11). Kinetic radiolabeling and pulse-chase studies suggest that these empty capsids and intermediate particles are precursors of mature virions (34, 35, 57). On the basis of these early studies, it is generally believed that the viral genome and core proteins are inserted into preassembled empty capsids. Studies of the protein composition of empty capsids and intermediate particles also support the precursor-product relationship between these particles and mature virions. Some of the proteins in the empty capsids and intermediate particles are scaffolding proteins such as the 100-kDa protein, which is required for hexon trimerization (40), while others are precursors of the capsid proteins in mature virions such as pVI, pVIII, and pIIIa. After encapsidation, a final maturation step is mediated by the viral protease, which cleaves the precursors to mature proteins in the capsids.

Adenovirus DNA packaging requires the packaging sequence located at the left end of the viral genome (nucleotides [nt] 194 to 358 in adenovirus type 5 [Ad5]) (28, 31). This region in Ad5 contains seven functionally redundant sequence elements called A repeats (21, 22). Most of the A repeats have a consensus motif that is important for DNA packaging (21, 22, 52, 53). A cellular factor called the P complex binds to these repeats, but its role in encapsidation is not yet understood (53). Several viral proteins have been shown to be involved in DNA packaging. Temperature-sensitive mutant viruses that express the L1 52/55-kDa and IIIa proteins, ts369 and ts112, respectively, accumulate intermediate particles containing only the left end of the viral genome when grown at the nonpermissive temperature (12, 30). By constructing an Ad5-derived mutant virus, pm8001, which does not express any L1 52/55-kDa protein, our group has previously demonstrated that the L1 52/55-kDa protein is required for viral DNA encapsidation: viral particles isolated from pm8001-infected cells contain no DNA (27).

Studies of the function of the IVa2 protein indicated that it also plays a role in viral DNA packaging. While the IVa2 protein was first identified as a transcriptional activator of the major late promoter (MLP) (38, 39, 59), we have demonstrated that the IVa2 protein binds to sequence motifs in the packaging sequence which are also found in the MLP (68, 69). This, along with the observation that the IVa2 and 52/55-kDa proteins interact in infected cells, indicated a possible role for the IVa2 protein in DNA encapsidation (27). Further evidence that the IVa2 protein is involved in DNA packaging came from an exploration of the specificity of DNA packaging among adenovirus serotypes. A chimeric virus containing the Ad7 genome except for the inverted terminal repeats and packaging sequence, which are from Ad5, can replicate its DNA and express its genes in 293 cells, but no infectious viruses are produced (69). However, 293 cells expressing the Ad5 IVa2 protein can support the growth of the chimeric virus, indicating that the IVa2 protein plays a role in viral DNA packaging and that a functional interaction between the IVa2 protein and the rest of the adenovirus packaging machinery is serotype specific.

Since the IVa2 protein is present in both empty capsids and mature virions (27, 67), is one of the virus core components (19), and has the properties described above, it is possible that it is involved in DNA packaging. To definitively establish such a role for the IVa2 protein, we have constructed a mutant virus that does not express the IVa2 protein. Analysis of the replication of this virus indicates that the IVa2 protein is required for both capsid assembly and DNA encapsidation. Moreover, limiting the level of IVa2 protein results in the production of abnormal viral particles. These results raise the possibility that adenovirus may be assembled by building the virion around the DNA core rather than by packaging DNA into preformed capsids.

MATERIALS AND METHODS

Cells and viruses.

293 cells are human embryonic kidney cells expressing Ad5 E1A and E1B proteins (24). A549 is a human lung cancer cell line (18). These cells were maintained in Dulbecco modified Eagle medium (DMEM) with 10% fetal bovine serum (FBS), 100 U of penicillin per ml, and 100 μg of streptomycin per ml. 293-IVa2 cells are 293 cells that stably express the Ad5 IVa2 protein and were maintained in DMEM with 10% FBS, 500 μg of G418 per ml, and 100 μg of hygromycin per ml (69). C7 cells, which are 293 cells that express the Ad5 DNA polymerase and preterminal proteins (1), were grown in DMEM with 10% FBS and 100 μg of hygromycin per ml. All cells were grown in a 5% CO₂ environment in a humidified 37°C incubator.

Wild-type Ad5 (from American Type Culture Collection) was propagated on 293 cells as described previously (23). The titers of the viruses used in this study were determined by a fluorescent focus assay. The virus to be titrated was diluted as serial l0-fold dilutions in medium containing 2% FBS. Confluent 293 cells were infected with the viral samples and incubated at 37°C for 2 h with gentle rocking every 20 min. The cells were further incubated in DMEM with 10% FBS for 28 h. Then the infected cells were washed two times with phosphate-buffered saline (PBS) and fixed with 50% methanol plus 50% acetone at room temperature for at least 10 min followed by two washes with PBS. Rabbit antibody against the L1 52/55-kDa protein was added to the cells at a 1:500 dilution in PBS, and the cells were incubated at room temperature for 40 min. The cells were washed twice with PBS to remove nonspecifically bound antibody, and fluorescein-conjugated goat anti-rabbit immunoglobulin G (Roche) diluted at 1:100 in PBS with 0.25% bovine serum albumin was added to the cells. The cells were incubated for 30 min, washed twice with PBS, and examined under a fluorescence microscope. The titer (fluorescent focus units [FFU]) was calculated on the basis of the average number of fluorescing cells per well at a dilution which allowed at least 200 positive cells per low-power field to be counted.

All infections to study the viral life cycle were performed at a multiplicity of infection (MOI) of 5 FFU/cell. Virus was allowed to adsorb for 2 h in DMEM with 2% FBS with gentle mixing every 15 min, followed by addition of DMEM with 10% FBS, and infected cells were harvested at the indicated times.

Construction of pTG3602-mIVa2.

The strategy for introducing point mutations in the IVa2 open reading frame (ORF) was similar to that described previously for the L1 52/55-kDa gene (27). First, the mutations were introduced by PCR into a 2.1-kb fragment (nt 4484 to 6617 in Ad5) containing the IVa2 gene (Fig. 1). Figure 1 shows the sequence of primer B (nt 5378 to 5402), which contains three point mutations (C to T). Primer C is complementary to primer B. These mutations generate stop codons at residues 17 and 19 of the IVa2 ORF and two restriction endonuclease sites (SpeI and AflII). The mutations do not change the amino acid sequence in the viral DNA polymerase ORF, which overlaps the IVa2 ORF. Fragment AC (nt 4484 to 5402) was amplified using primers A (nt 4484 to 4505, CAGAACCACCAGCACAGTGTA) and C. Fragment BD (nt 5378 to 6617) was amplified using primers B and D (nt 6617 to 6596, AGATAGACTACTTCGACGCGC). Then fragment AD, containing the point mutations, was amplified using primers A and D with equal amounts of mixed fragments AC and BD as templates. Fragment AD was gel purified, and the presence of the point mutations was confirmed by digesting the fragment with SpeI or AflII. These point mutations were then introduced into pTG3602, which contains the full-length Ad5 genome, by homologous recombination in Escherichia coli BJ5183 (9). To perform this recombination, a single endonuclease digestion site that is adjacent to the desired crossover site is required in pTG3602. There are two Bst1077I sites in the Ad5 genome, one at nt 5766 and the other at nt 29012. The second site (nt 29012) was temporarily removed by excising an internal NdeI fragment (nt 19549 to 31089) in pTG3602 and religating to make pTG3602-NdeI. Then, 1.5 fmol of Bst1077I-digested pTG3602-NdeI and 15 fmol of fragment AD were cotransformed into BJ5183. Twelve colonies were picked, and miniprep DNA was isolated and transformed into DH5α cells, which give a better yield of DNA. Colonies were picked, and DNA was extracted and digested with SpeI to screen for the presence of the mutations. DNA from positive clones was digested with multiple restriction enzymes to ensure the structure of the rest of the genome. One clone, named pTG3602-NdeI-mIVa2, was digested with NdeI and ligated with the previously removed NdeI fragment (nt 19549 to 31089) to generate pTG3602-mIVa2. The genomic organization of pTG3602-mIVa2 is the same as wild-type Ad5 except for the IVa2 point mutations, as determined by digestion with multiple restriction enzymes.

FIG. 1.

IVa2 mutations in pm8002. The black boxes represent the two exons of the IVa2 gene. The intron (INT) and the positions of the primers (A to D) described in Materials and Methods are indicated. The three point mutations, which generated stop codons at amino acids 17 and 19 (underlined), and two restriction enzyme sites, SpeI and AflII, are shown.

pm8002.

To generate a virus with the IVa2 point mutations, 293 or 293-IVa2 cells were transfected with 10 μg of pTG3602-mIVa2 that had been digested with PacI to release the viral genome. The production of virus after transfection was monitored by the appearance of cytopathic effect (CPE). The transfected 293 cells did not develop CPE, whereas the 293-IVa2 cells developed CPE 10 to 14 days after transfection. These cells were harvested and resuspended in DMEM with 2% FBS, and a lysate was prepared by freezing and thawing the cells three times. The lysate was plaqued on 293-IVa2 cells, and four isolated plaques were picked.

Isolation of cellular and nuclear viral DNA.

Low-molecular-weight DNA was extracted from infected cells in 60-mm-diameter dishes by lysing the cells in a solution containing 0.6% sodium dodecyl sulfate and 10 mM EDTA as described previously (27, 33). The extracted DNA from one 60-mm-diameter dish was resuspended in 50 μl of TE (10 mM Tris, 1 mM EDTA [pH 8]). For viral DNA replication assays, to control for possible loss of DNA during the DNA extraction process, 100 μg of KpnI-digested pcDNA-TripIVa2 (69) was added to the cell lysates prior to extraction.

To isolate nuclear viral DNA, infected cells in 60-mm-diameter dishes were harvested and washed twice with ice-cold PBS. The cells were resuspended in ice-cold Iso-Hi pH buffer (10 mM Tris [pH 8.4], 140 mM NaCl, 1.5 mM MgCl₂). One-tenth volume of 5% NP-40 in Iso-Hi pH buffer was added to the suspension, and the sample was incubated on ice for 5 min. The nuclei were pelleted at 500 × g in a microcentrifuge for 5 min at 4°C and washed twice with ice-cold PBS prior to low-molecular-weight DNA isolation as described above. The DNA extracted from one 60-mm-diameter dish was resuspended in 50 μl of TE.

To assay viral DNA by semiquantitative PCR, 2 μl of the 50 μl of undiluted or 1:10 diluted cellular or nuclear viral DNA was used as a template for PCR. The primers used in these PCRs were primers A and D.

Purification of viral particles.

Viral particles from cells infected with pm8002 or wild-type Ad5 were purified as described previously (27). For each infection, 10 15-cm-diameter tissue culture dishes of infected cells were harvested after 48 h, centrifuged at 250 × g for 10 min, and washed once with PBS. The cells were then resuspended in 15 ml of 10 mM Tris (pH 8.0) and frozen and thawed three times. Cell debris was removed by centrifugation at 1,500 × g for 15 min. The supernatant was layered onto a 1.20- and 1.45-g/cm³ CsCl step gradient and centrifuged at 72,000 × g for 2 h at 20°C. The virus band at the interface of the two CsCl layers was collected, diluted with an equal volume of 10 mM Tris (pH 8.0), layered onto a preformed continuous CsCl gradient (1.20 to 1.45 g/cm³), and centrifuged at 72,000 × g for 16 h at 20°C. Fractions were collected from the bottom of the centrifuge tube. Buoyant density and spectrophotometric absorbance at 260 nm (A₂₆₀) of the fractions were measured.

Transmission electron microscopy.

Infected 293 or 293-IVa2 cells were analyzed by electron microscopy. At 24 h postinfection, cells were washed twice with cold DMEM without serum and fixed with 5% glutaraldehyde for 1 h. The samples were then processed and examined in the Microscopy and Image Analysis Laboratory at the University of Michigan.

Southern blot and immunoblot analyses.

For Southern blots, DNA samples from infected cells were digested with KpnI and SpeI, loaded on a 0.8% agarose gel for electrophoresis, and transferred to a GeneScreen Plus hybridization membrane (NEN Life Science Products, Inc., Boston, Mass.). The ³²P-labeled probe was generated by the Random Primer Labeling kit (Life Technologies Inc., Gaithersburg, Md.) using pTG3602. The hybridization procedure was described previously (69). Immunoblotting was also performed as previously described (29). The rabbit antibody against the L1 52/55-kDa protein was described previously (27), the rabbit antihexon antibody was obtained from Doug Brough (GenVec), and the monoclonal antibody against the 72,000-molecular-weight DNA binding protein (72-kDa DBP) is from Arnie Levine (Rockefeller University).

Preparation of anti-IVa2 antiserum.

Anti-IVa2 mouse serum was generated by immunizing mice with IVa2 protein purified from bacteria. The IVa2 cDNA was cloned into a glutathione S-transferase (GST) fusion vector, pGEX-5X-3 (Pharmacia). The GST-IVa2 fusion protein was expressed in E. coli and affinity purified with glutathione-Sepharose 4B. Factor Xa treatment was used to release the IVa2 protein from the column. The IVa2 protein was then gel purified prior to immunization. Immunization of mice was performed by the Hybridoma Core at the University of Michigan.

Marker rescue assay.

293 cells were cotransfected with PacI-digested pTG3602-mIVa2 and a PCR fragment (nt 3871 to 5943) that contains the wild-type IVa2 gene. This gene was amplified from wild-type Ad5 by using primers CGCGAAGCTTGTGCAGCTTCCCGTTCATC and GCGCAGATCTCGACATGTGTCTTCACACC. CPE was detected in the cotransfected cells after 10 days. The rescued virus was plaque purified, amplified, and titrated on 293 cells.

RESULTS

Generation of pm8002.

Transfection of the PacI-released mutant viral genome from pTG3602-mIVa2 into 293 cells did not result in production of infectious virus, indicating that the mutant virus was not able to grow in the absence of the IVa2 protein. In contrast, infectious viruses were produced in transfected 293-IVa2 cells. To confirm that the virus from the 293-IVa2 cells contained the mutations in the IVa2 gene, viral DNA from four individual plaque-purified viruses was extracted and used as a template for PCR with primers A and D. The PCR products, which span the IVa2 gene, were digested with SpeI or AflII. Only DNA with the point mutations can be digested by these two enzymes. The DNA fragments amplified from all four plaques were digested by both SpeI and AflII (data not shown). The virus from one plaque was amplified in 293-IVa2 cells and used for the rest of the studies.

Gene expression of pm8002.

To ensure that the mutations prevented the expression of the IVa2 protein, A549 cells were infected with pm8002 or wild-type Ad5 at an MOI of 5 FFU/cell. Cell lysates were prepared at different times after infection, and immunoblots were used to detect the IVa2 protein (Fig. 2). The IVa2 protein was detectable at 12 h postinfection in wild-type Ad5-infected A549 cells, and the amount of expression increased over time. In pm8002-infected A549 cells, full-length IVa2 protein was not detected even 24 h after infection. A small amount of a protein that migrated faster than the IVa2 protein was detected 24 h postinfection. The source of this protein, which was also present in wild-type lysates, is not known, but the fact that it is present only in infected cell extracts and is recognized by the anti-IVa2 serum indicates that it might be a truncated form of the IVa2 protein.

FIG. 2.

Gene expression of pm8002. A549 cells were infected with pm8002 or wild-type Ad5 at an MOI of 5. Protein lysates were collected at the indicated times (in hours) after infection. Immunoblotting was performed with antibodies against IVa2, the L1 52/55-kDa protein, 72-kDa DBP, hexon, and E1A, as indicated to the right of the blots.

Since the IVa2 protein has been reported to be a transcriptional activator of the MLP (38, 39, 59), we expected to see a decrease in the expression of late viral proteins in the absence of the IVa2 protein. The mutant virus was able to express the hexon protein, but its expression was delayed by about 6 h from that of the wild-type virus (Fig. 2). We also expected that early gene expression would not be affected. Surprisingly, we found that expression of the L1 52/55-kDa protein and 72-kDa DBP was also delayed (Fig. 2). Moreover, this delay was also seen in 293 and 293-IVa2 cells (data not shown), indicating that it was independent of the onset of expression of E1A, which is required for expression of the L1 52/55-kDa protein and 72-kDa DBP during the early phase of infection. In fact, expression of the E1A protein in A549 cells was also delayed by about 6 h (Fig. 2). These results indicate that the lag in gene expression was unlikely to be due to a direct effect of the loss of IVa2 transcriptional activity, since both early and late gene expression was affected.

DNA replication of pm8002.

Since the expression of the 72-kDa DBP was delayed, we assumed that DNA replication of the mutant virus would also be delayed. 293 cells were infected with pm8002 or wild-type Ad5 and harvested at various times after infection. Viral DNA was extracted and analyzed by Southern blotting (Fig. 3A). DNA replication was detected at 12 h postinfection in Ad5-infected 293 cells, and the amount of viral DNA increased over time. In pm8002-infected 293 cells, DNA replication was not detected until 18 h postinfection, indicating that the mutant virus was able to replicate its DNA but with a delay that mirrored early gene expression. Although the mutations in the IVa2 ORF were designed not to affect the DNA polymerase gene, we wished to ensure that the effect on replication was not due to changes in DNA polymerase expression as a result of the presumably silent mutations. For this purpose, C7 cells, which express DNA polymerase and have been shown to complement the growth of viruses that do not express this enzyme, were used. We found that C7 cells did not overcome the delay in DNA replication (Fig. 3B).

FIG. 3.

DNA replication of pm8002. 293 (A) or C7 (B) cells were infected with pm8002 or wild-type Ad5 and harvested at the indicated times (in hours) after infection. Viral DNA was extracted, digested with KpnI and SpeI, and analyzed by Southern blotting using an Ad5 probe. In both panels, the black arrowhead and arrows indicate wild-type Ad5- and pm8002-specific bands, respectively. In panel B, the white arrow to the left of the blot points to the control band resulting from inclusion of digested plasmid DNA as an internal standard prior to DNA extraction (Materials and Methods).

pm8002 DNA trafficking to the nucleus is delayed.

Since E1A expression was delayed, we determined whether pm8002 was trafficking to the nucleus efficiently. To measure the amount of DNA at very early times in pm8002- or Ad5-infected cells, viral DNA was extracted from cells 4 h postinfection and analyzed in a semiquantitative PCR. Similar amounts of amplified DNA were obtained from the viral DNA from both pm8002- and Ad5-infected cells (Fig. 4A), indicating that the total cell-associated viral DNA at 4 h after infection was not different in the two viruses. We then asked if similar amounts of DNA could be found in the nuclei of the infected cells. Nuclei were isolated from cells at 1 and 4 h after infection, viral DNA was extracted, and PCR was performed (Fig. 4B). The amount of amplified DNA from pm8002-infected nuclei at both times was reproducibly less than that from the Ad5-infected nuclei. These results indicate that the transport of viral DNA to the nucleus was delayed in the pm8002-infected cells, thereby explaining the late onset of E1A expression.

FIG. 4.

pm8002 trafficking to the nucleus. Viral DNA was extracted from pm8002- or Ad5-infected 293 cells (A) or purified nuclei (B) at 1 or 4 h after infection. PCR was performed to quantify the DNA using primers for the IVa2 gene. Two different dilutions (1:1 and 1:10) of the template DNA were used in panel A, as described in Materials and Methods.

Analysis of viral particles in pm8002-infected cells.

To determine whether the IVa2 protein is required for DNA encapsidation, viral particles were examined. 293 or 293-IVa2 cells were infected with pm8002 or Ad5 at an MOI of 5 FFU/cell, and 48 h later, cells were harvested and viral particles were separated on CsCl gradients. The buoyant density and spectrophotometric A₂₆₀ of the fractions were measured (Fig. 5). There was a major peak at a density of 1.34 g/cm³ from Ad5-infected 293 cells, representing mature virions. The titer of the virus from the peak was 2 × 10¹¹ FFU/ml (Table 1). In the gradients from pm8002-infected 293 cells, no absorbance peak at 1.34 g/cm³ was detected. However, virus with a titer of 1.4 × 10⁶ FFU/ml was recovered from this fraction. This might simply be carryover of the initial inoculating virus, or it might represent a very small amount of newly assembled virus. In either case, the results indicate that the mutant virus was severely defective at assembling mature virions. In the gradients from pm8002-infected 293-IVa2 cells, there was a peak at the position of mature virions along with a broad shoulder of less dense particles, although the height of the peak was much smaller (titer = 1.2 × 10⁹ FFU/ml) than that from the Ad5 fractions, and the particle/FFU ratio was higher (35 for Ad5 and 200 for pm8002 [Table 1]). This suggested that the pm8002-infected 293-IVa2 cells were able to produce mature viral particles but to a lesser extent than the wild-type virus, and the infectivity of these particles was decreased. The amount of IVa2 in the pm8002 peak is also decreased about fivefold (Fig. 5B). This could indicate that only one-fifth of the particles contain a full complement of IVa2 protein with the rest containing none. Such an interpretation is consistent with the particle/FFU ratio. The material in the shoulder also was infectious, with a much higher particle/FFU ratio, but we did not measure the IVa2 in these particles (Table 1).

FIG. 5.

CsCl gradient analysis of viral particles. 293 cells or 293-IVa2 cells were infected with pm8002 or Ad5 at an MOI of 5. CsCl gradient centrifugation was performed 48 h postinfection, and fractions were collected from the bottom of the tube. (A) The density and absorbance of each fraction were measured. Note the difference in scale of the y axis between the top graph and the other two. (B) The amount of IVa2 protein in pm8002 viral particles was analyzed by immunoblotting with anti-IVa2 antibodies. Lane 1, 5 × 10⁹ viral particles from fraction 12 of the pm8002/293-IVa2 gradient; lane 5, 5 × 10⁹ viral particles from fraction 14 of the Ad5/293 gradient; lanes 2 to 4, 27-, 9-, and 3-fold dilutions of the sample in lane 5, respectively.

TABLE 1.

Analysis of CsCl gradient fractions

Cell and fraction	Virus titer (FFU/ml)	Particle/FFU ratio
Ad5-infected 293
1	NDa
14b	2 × 1011	35
pm8002-infected 293
1	ND
11b	1.4 × 106	9,285
15	ND
pm8002-infected 293-IVa2
1	ND
12b	1.2 × 109	200
14	1.6 × 108	885
16	2 × 107	2,500
18	1.4 × 106	64,285
21	ND

^aND, not detected.

^bFractions with a density of 1.34 g/cm³.

To visualize the viral particles in infected cells, we examined cells using electron microscopy (Fig. 6). Mock-infected 293 cells were used as a negative control. Ad5-infected cells contained icosahedral, dark- or light-staining particles, which are thought to represent mature virions or assembly intermediates, respectively. pm8002-infected 293-IVa2 cells contained viral particles, but most of these particles differed from the particles in Ad5-infected cells in size and shape. Due to their abundance, we believe that these are the lighter density particles detected on the CsCl gradients. Of most interest is the finding that pm8002-infected 293 cells contained no detectable viral particles. Taken together, these results indicated that the IVa2 protein not only plays a role in encapsidation but is also required for capsid assembly and that 293-IVa2 cells only partially complement the mutant virus.

FIG. 6.

Electron microscopic analysis of virus-infected cells. A mock-infected 293 cell (A), an Ad5-infected 293 cell (B), a pm8002-infected 293 cell (C), and a pm8002-infected 293-IVa2 cell (D) are shown. nm, nuclear membrane. Bars, 0.3 μm.

Since the particles produced in the pm8002-infected 293-IVa2 cells demonstrated a delay in trafficking to the nucleus and displayed abnormal sizes and shapes, we wished to examine if the deficiency in IVa2 protein resulted in particles that were less stable than wild-type virions. We incubated viral lysates from pm8002-infected 293-IVa2 cells or Ad5-infected 293 cells at 42°C for various periods of time and then determined the titer of the virus by fluorescent focus assay. The mutant virus had reduced stability relative to wild-type Ad5 (Fig. 7).

FIG. 7.

pm8002 is not stable at 42°C. Viral lysates from pm8002-infected 293-IVa2 cells or Ad5-infected 293 cells were incubated at 42°C for various periods of time after which they were assayed on 293 cells in a fluorescent focus assay. The results are presented as percentages of the virus titer before incubation at 42°C.

Marker rescue.

Since the 293-IVa2 cells did not fully complement the mutant virus, it was possible that there were mutations in the viral genome other than those in the IVa2 gene. We therefore determined whether a wild-type IVa2 DNA fragment could rescue the mutant virus in a cotransfection experiment. The mutant virus could be fully rescued: the IVa2 protein was expressed from the rescued virus at wild-type levels, and there was no delay in the expression of the E1A, 72-kDa DBP, and hexon proteins (data not shown). These results indicated that no other mutations were present in the pm8002 genome.

DISCUSSION

We have previously postulated that the adenovirus IVa2 protein plays a role in viral DNA encapsidation. The IVa2 protein interacts with motifs in the packaging sequence (68) that are critical for viral DNA packaging (52, 53), and the Ad5 IVa2 protein is required for the growth of a chimeric virus which contains the Ad7 genome with the Ad5 inverted terminal repeats and packaging sequence (69). In this study, we dissected the function of the IVa2 protein further by analyzing an IVa2 null mutant virus, pm8002. In 293 cells, the mutant virus can express its early and late genes and replicate its DNA. However, these cells produce no viral particles detectable by either CsCl gradient purification or electron microscopy analysis, strongly indicating that the IVa2 protein is required for assembly of both empty capsids and mature virions. This unexpected function of the IVa2 protein in capsid assembly is consistent with the fact that the IVa2 protein is a component of the virus core complex, empty capsids, and mature virions (19, 27, 67). While an apparently truncated form of the IVa2 protein is produced from pm8002, this protein seems to be nonfunctional.

We were able to generate infectious pm8002 viral particles in the 293-IVa2 helper cell line, but these cells did not fully complement the IVa2 mutation. The titer of infectious mutant virus produced by this cell line was 100-fold lower than that from a wild-type virus infection, and many of the viral particles produced in this cell line were not mature virions, as they exhibited sizes and shapes different from those of the typical icosahedral structure. These abnormal particles could not traffic efficiently to the nucleus and were not as stable as wild-type virus at 42°C. Wild-type Ad5 grew normally in these cells (data not shown), indicating that they are not inhibitory for adenovirus growth. One possible reason why the 293-IVa2 cell line could not fully complement the mutant virus is that there may have been mutations in the viral genome other than the intended mutations in the IVa2 gene; however, this possibility was ruled out by the marker rescue assay results. Another possibility is that the amount of the IVa2 protein expressed in the helper cell line, which is about 20-fold less than that in wild-type virus-infected cells at 24 h postinfection (data not shown), is not enough to fully complement the mutant virus. There are precedents for believing that the virus needs to produce a threshold amount of the IVa2 protein for efficient assembly. First, it has been shown that cotransfection of Ad5 DNA with a plasmid containing packaging domains, which can compete for packaging factors, results in a decrease in viral yield (22). Second, while certain mutant viruses containing suboptimal packaging sequences can grow to wild-type titers on their own, in the presence of wild-type virus their packaging is dramatically reduced, again arguing for a limiting factor(s) in the reaction (52). Since the IVa2 protein binds to repeated motifs in the packaging sequence (68), it may be the limiting molecule in both situations. Also, the packaging sequence overlaps the E1A enhancer, which contains binding sites for cellular factors such as EF-1A and E2F (6, 32, 36). These cellular factors, as well as the P complex (53), might compete with the IVa2 protein for binding to the packaging sequence. Overall, we conclude that the level of IVa2 expression may be critical for efficient completion of the adenovirus life cycle.

The IVa2 protein has been described as a transcriptional activator of the MLP (38, 39, 59). We find that the IVa2 protein is not required for optimal levels of late viral gene expression. The delay in late gene expression in pm8002 appears to be due mainly to the delay in early protein expression, which results from the inefficient trafficking of the partially complemented abnormal virus particles to the nucleus. It is not clear why the delivery of DNA to the nucleus by the mutant virus particles is delayed. In order to transport DNA into the nucleus, the capsid needs to go through several steps: binding to the receptor, coxsackievirus and adnovirus receptor (3), internalization (64-66), endosomal escape (10), and binding to the nuclear pore complex receptor (25, 60). A problem with any one of these steps could cause the phenotype we detect. It is possible that the partially complemented particles disassemble prematurely in the cell, since these particles are less stable than wild-type virions. We also cannot rule out the possibility that the IVa2 protein transactivates E1A transcription as the packaging sequence overlaps the E1A enhancer. Such an activity would be similar to that of the herpes simplex virus (HSV) VP16 protein, which is a virion component that activates expression of HSV immediate-early genes (2, 7).

It has been thought that the adenovirus genome and core proteins are inserted into preformed empty capsids. Many bacteriophages have been shown to package their DNA into procapsids (8, 14, 17, 26, 54). In mammalian viruses, HSV has been described to be using a pathway similar to that of phage λ to package its DNA (41, 42, 56). This model recently received significant support with the identification of a portal protein on the HSV capsid (43). For adenovirus, there is a great deal of evidence that supports a similar model. Incomplete particles containing variable lengths of the left end of the viral genome accumulate in cells infected with the temperature-sensitive mutant viruses ts369 and ts112 at the nonpermissive temperature (12, 30). The behavior of ts369, whose mutation maps to the L1 52/55-kDa gene, and our finding that viral particles isolated from a L1 52/55-kDa null mutant virus-infected cells contain no viral DNA (27) imply that capsids are constructed first. This hypothesis is also supported by the results of older kinetic radiolabeling and pulse-chase studies. Radiolabeled amino acids are first found in incomplete particles and then in mature virions 60 to 80 min later (57), indicating that the former particles are assembled first. Furthermore, the radiolabel in incomplete particles can be chased into mature virions, suggesting a precursor-product relationship (57).

Since empty capsids are icosahedral structures, if the viral DNA is inserted into these structures, we would expect a population of homogeneous icosahedral viral particles in cells infected with pm8002 if the only function of the IVa2 protein were to effect DNA encapsidation. The absence of any particles in pm8002-infected 293 cells, along with the abnormal particles that we detect in 293-IVa2 cells, however, may suggest that adenovirus is assembled through a different mechanism in which the viral structural proteins are built around the nucleoprotein core in an IVa2-dependent manner. The process may be inefficient in the 293-IVa2 cells due to limiting IVa2 protein levels. Heterogeneous populations of viral particles have been isolated from wild-type virus-infected cells. Some of these intermediate particles must be purified by a mild method such as Ficoll gradients, as they are not stable enough to survive CsCl gradient centrifugation (15). We would argue that these intermediates are not abundant in a wild-type virus infection because the levels of IVa2 are sufficient to complete the assembly process. Edvardsson et al. have also shown that two temperature-sensitive mutants, ts19 (pX phosphorylation defective) and ts58 (IIIa defective), accumulate similar intermediates at the nonpermissive temperature, a fraction of which are processed into mature virions when the infected cells are shifted to the permissive temperature (16). Therefore, it is possible that the heterogeneous intermediates defined by Edvardsson and colleagues are the true precursors to mature virions and that empty capsids are dead end products.

Poliovirus is one of the mammalian viruses that may assemble its structural proteins around the genome, which is RNA in this case. There is evidence that 14S subunits, which are pentamers of VP0-VP1-VP3, are the key assembly intermediates in poliovirus assembly. The 14S subunits are always present in infected cells (48), can be chased into mature virions (49, 50), and are associated with viral RNA (44, 46). Furthermore, in a cell-free system, the 14S subunits can package the viral genome into virions, while procapsids cannot under the same conditions, indicating that the procapsids are either dead end products or a reservoir for the 14S subunits (4, 63). Among the DNA viruses, simian virus 40 is thought to assemble its capsid around its circular DNA genome. Most recently, Gordon-Shaag et al. have shown that a cellular factor, Sp1, recruits two of the simian virus 40 capsid proteins, VP2 and VP3, to the viral packaging signal, ses, to form a packaging center around which VP1 assembles (20).

Additional support for a model in which adenovirus assembles capsids around the DNA comes from studies on adenovirus DNA cores. It has been shown that the core protein pVII has functional features in common with cellular histones (for example, condensing the DNA [37, 58]) and that intranuclear viral DNA is organized in a manner similar to that of viral chromatin in mature virions (5, 51). Furthermore, the intranuclear chromatin core structure becomes dominant during the late stage of the viral infection (13). Thus, cores can form without prior insertion into capsids. In such a scenario, a possible explanation of the pulse-chase experiments is that the formation of empty capsids is a reversible process in which the structural proteins in the empty capsids may dissociate and reassemble around the DNA core. Alternatively, empty capsids may not be part of the normal assembly process. These findings, together with our current results, support a model in which the DNA core complex forms first and then the structural proteins are assembled around it to produce mature virions.

Finally, we cannot rule out the possibility that the defective particles in the IVa2 mutant virus-infected helper cell line are formed only in this special situation and do not reflect the normal viral assembly process. If this were the case, we would suggest that the IVa2 protein plays two distinct roles in adenovirus assembly, the first being assembly of empty capsids and the second being encapsidation of DNA into those capsids. More-detailed study will be required to distinguish the two possible assembly pathways discussed herein.