The Papillomavirus Minor Capsid Protein, L2, Induces Localization of the Major Capsid Protein, L1, and the Viral Transcription/Replication Protein, E2, to PML Oncogenic Domains

Abbreviations used in this paper.

Ad5, adenovirus type 5; BPV, bovine papillomavirus; FITC, fluorescein isothiocyanate; HPV, human papillomavirus; HSV, herpes simplex virus; IgG, immunoglobulin G; NIH, National Institutes of Health; PBS, phosphate-buffered saline; PML, promyelocytic leukemia protein; POD, PML oncogenic domain; SFV, Semliki Forest virus; SV40, simian virus 40; VLP, virus-like particle.

Antibodies.

Monoclonal antibody B201, directed against the BPV E2 protein, and polyclonal antiserum 150-1, which recognizes the BPV E1 protein, were provided by Elliot Androphy (New England Medical Center, Tufts University School of Medicine). Monoclonal antibody 5B6, which recognizes the BPV L1 capsid protein, and rabbit polyclonal antiserum 17/28, raised against the full-length BPV L2 capsid protein, were generated in this laboratory and have been described previously (46). Monoclonal antibody 6A8, directed against the BPV L2 protein, was provided by A. Bennett Jenson (Georgetown University) (28). The antibody against SC35 was purchased from Sigma Chemicals (St. Louis, Mo.). The anti-PML antibody, 5E10, was generated by R. van Driel (University of Amsterdam) and was a kind gift of Louis Staudt (National Cancer Institute) (50). FITC-conjugated goat anti-mouse IgG and Texas red-conjugated goat anti-rabbit IgG were purchased from Jackson Immunoresearch (West Grove, Pa.).

Cell lines.

BPHE-1 cells, obtained from Andrew Lewis Jr. (NIH), were grown in Dulbecco’s modified Eagle’s medium supplemented with antibiotics and 10% fetal calf serum (57). BHK-21 cells were grown in Glascow’s medium supplemented with 10% tryptose phosphate broth, antibiotics, nonessential amino acids, HEPES, and 5% fetal calf serum. For microscopic analyses, the cells were seeded onto acid-washed no. 01 coverslips in 24-well plates at a density of 10⁵ cells/well and cultured overnight.

Recombinant SFV expression system.

The production of recombinant SFV RNAs and replication-defective virus expressing the BPV L1 or L2 capsid protein and the SFV infection protocols have been described previously (45). BPV E1 and E2 were cloned into the BamHI site of pSFV-1 as PCR products amplified from the BPV genome (primers for E1, 5′ ccgctggatccgcaccatggcaaacgataaaggtagc and 3′ gcggtggatccgatcttgcaacttatcactac; primers for E2, 5′ ccgctggatccgcaccatggagacagcatgcgaacg and 3′ gcggtggatccgaagaaaaggcaatggcagtg [boldface indicates restriction sites and start sites]), and recombinant viruses expressing each gene were generated as described for L1 and L2. For infection of cells, high-titer recombinant SFV stock was treated with 500 μg of chymotrypsin A4 per ml on ice for 30 min and then aprotinin was added to 500 μg/ml for an additional 10 min. The activated virus was diluted 100-fold in Dulbecco’s PBS with calcium and magnesium and added to cells in 24-well plates. After 60 min at 37°C, virus-containing medium was removed and replaced with the normal growth medium supplemented with 100 mM KCl for the remainder of the infection to maintain cellular protein expression. Infections were allowed to continue for 5 to 6 h prior to cell fixation and immunolocalization. Although SFV infection will induce cell death in 48 h, the morphology of the infected cells was not visibly altered at this early time point.

Immunofluorescence staining.

The cells were washed three times with cold PBS (pH 7.4), fixed by a 10-min incubation at room temperature with 1.0% paraformaldehyde diluted in PBS, and washed three times with PBS–200 mM glycine. They were then incubated with primary antibody diluted in PBS–0.1% polyoxyethylene 20 cetyl ether (Brij 58; Sigma Chemicals) and incubated at 4°C. Polyclonal antisera were used at a dilution of 1/1,000. Monoclonal antibodies used as hybridoma supernatants were diluted 1/100. Purified antibodies were used at a concentration of 5 μg/ml. For double immunofluorescence staining, the primary antibodies were incubated in unison. After incubation, the coverslips were washed three times with PBS–0.1% Brij. Secondary antibodies were diluted to 5 μg/ml in PBS–0.1% Brij, and incubation was performed at 4°C. After this incubation, the cells were washed thoroughly in PBS–0.1% Brij and inverted onto Fluoromount-G mounting solution (Southern Biotechnology Associates, Birmingham, Ala.) on a glass slide. Fluorescence was examined with a Bio-Rad MRC 1024 laser scanning confocal system attached to a Zeiss Axioplan microscope. All the images were acquired with a Zeiss 63x N.A. 1.4 Planapo objective, using the photon-counting mode. The use of control coverslips established that fluorescence in the green and red channels was not overlapping and that antibody binding was specific for the intended antigen. The images were collaged and subjected to scale adjustment with Adobe Photoshop software.

Subnuclear localization of BPV capsid proteins.

BPHE-1 is a hamster fibroblast cell line that is latently infected with multiple copies of autonomously replicating BPV genomes and expresses the nonstructural viral proteins (57). We used the SFV expression system to introduce the L2 minor capsid protein into BPHE-1 cells and localized the L2 protein by immunofluorescence staining and laser scanning confocal microscopy. Figure 1A and D, which shows the typical distribution of L2 6 h after SFV infection, indicates that the protein was displayed in a distinct intranuclear punctate pattern.

To rule out the possibility that this distribution depended on the BPV components in the BPHE-1 cells, L2 was expressed, via the SFV vector, in cells that did not harbor papillomavirus sequences. A similar punctate nuclear pattern of L2 staining was also observed in other cell types, including COS-7, BHK-21, and the human fibroblast cell line 1634 (see Fig. 5A; data not shown). Therefore, this distinct L2 localization is dependent only upon cellular factors and appears to be independent of cell lineage. To determine if this localization was a common feature of papillomavirus L2, the distribution of the HPV16 L2 protein, expressed via an SFV vector, in these cell lines was also examined. The pattern with HPV16 L2 protein was similar to that seen with BPV L2 (data not shown), strongly suggesting that this localization is characteristic of papillomavirus L2.

L2-containing punctate structures are PODs.

To identify the nuclear domains in which the BPV L2 protein localized, we performed double-staining experiments with a number of described nuclear proteins and L2. We found no colocalization of the L2 protein with coiled bodies, the retinoblastoma protein p53, or the splicing factor SC35. Although the staining pattern seen with the anti-SC35 antibody was similar to that seen with the anti-L2 antibody (Fig. 1A and B), it was evident from the merged image that these regions were exclusive (Fig. 1C). However, when we compared the distribution of the L2 protein with that of anti-PML protein staining, we observed a nearly complete overlap in protein distribution (Fig. 1D to F).

The PML protein is a putative growth suppressor gene product that localizes in subnuclear organelles termed PODs (6, 14). The PML distribution appeared to be unaffected by the expression of the L2 protein, and the localization of L2 in the PODs was unrelated to the level of L2 in the cell. This is seen in Fig. 1D, where cells expressing high, intermediate, and low levels of L2 can be compared. All the cells expressing L2 showed a similar punctate distribution, in which L2 colocalized with PML in every cell. Therefore, it is unlikely that this colocalization is an artifact of overexpression. It is also unlikely that localization of L2 to PODs is due to expression via defective SFV, since neither morphological changes nor apoptosis (as evidenced by DNA fragmentation) was detected at the 5- to 6-h postinfection time point at which the analysis was conducted (data not shown). In addition, SFV infection did not induce POD localization of other proteins, independent of L2 (see below).

L2 redirects L1 to PODs.

Since L1 and L2 coassemble into capsids, we sought to determine whether L1 might display a nuclear staining pattern similar to that of L2. However, when L1 was expressed in BPHE-1 cells, its distribution protein differed markedly from that of L2. L1 was present in a nuclear pattern that varied from a diffuse to a slightly speckled arrangement with nucleolar exclusion (Fig. 2A).

This result led us to explore the possibility that the subcellular distribution of L1 protein is affected by coexpression of L2. Therefore, we coinfected BPHE-1 cells with recombinant L1-SFV and recombinant L2-SFV, which are the conditions that lead to the formation of infectious BPV in BPHE-1 cells. The L1 staining, which was collected in the green channel, is shown in Fig. 2B and E. This pattern was dramatically altered from the diffuse nuclear pattern seen after L1 SFV infection alone. The L2 staining pattern in the coinfected cells was consistent with the distribution of L2 observed in the absence of L1 (Fig. 2C and F). The distributions of L1 and L2 overlapped substantially in the merge of the two images (Fig. 2D and G). In general, L1 did not appear as tightly coalesced as L2 (Fig. 2D). However in some cells, L1 almost completely colocalized with L2 (Fig. 2G), while in other cells, we have observed L1 mostly surrounding rather than overlapping the L2 domain. This variability may be due to differences in the kinetics of the infection of individual cells or may reflect different stages in L1 relocation or virion assembly. We conclude that L2 induced the redirection of a substantial proportion of L1 to PODs.

L2 induces colocalization of E2.

We next examined the effect of expression of the BPV capsid proteins on the distribution of the nonstructural viral protein E2, which is involved in viral genome replication and viral transcription (8, 47, 54). In BPHE-1 cells, E2 was detected as a nuclear protein with a diffuse distribution (Fig. 3A). There was no apparent effect on the localization of this protein when the L1 capsid protein was expressed in these cells (Fig. 3B). In contrast, L2 expression shifted E2 into punctate regions similar to those observed with the anti-L2 staining shown in Fig. 1. Figure 3C shows representative E2 immunostaining in BPHE-1 cells that have been infected with recombinant L2-SFV. The levels of E2 often decreased substantially during recombinant SFV infection, presumably due to the well-documented inhibition of host protein synthesis by SFV, although this did not interfere with determining the localization of E2 (49). This effect is partially due to interference with the Na⁺K⁺ transporter by SFV (4, 18). A decrease in E2 was also observed in control infections with unrelated SFV recombinants (data not shown). Infection in the presence of 100 mM KCl helped counteract this problem, but nevertheless, as seen in Fig. 4B and C, the level of immunoreactive E2 was low in a number of the SFV-infected cells.

To determine if L2 induced the redistribution of the E2 protein into the L2-staining PODs, we performed double staining of the BPHE-1 cells after infection with L2-SFV. Figure 4A shows the E2 staining in these cells as detected by fluorescence in the green channel. Most of the cells show the diffusely distributed nuclear pattern. However, the cells indicated by the arrows demonstrate the relocation of E2 into the punctate pattern. Figure 4B shows the anti-L2 staining of these cells as detected in the red channel. All of the L2-expressing cells showed the punctate pattern of staining. Figure 4C is the merge of the red and green channels shown in Fig. 4A and B. The coincidence of the E2 and L2 staining is striking in the infected cells that maintain detectable levels of E2.

L2 is sufficient to redistribute full-length E2.

BPV-transformed cells with autonomously replicating genomes express three forms of the E2 protein: a full-length 48-kDa form that functions in genome replication and transcriptional transactivation and two smaller forms which act as repressors of viral transcription (25, 32, 40). The antibody used in the immunofluorescence studies recognizes an epitope in the C-terminal DNA binding domain common to all three proteins and would not distinguish among them. Another feature of the BPHE-1 cells is that an unknown proportion of E2 molecules are bound to the viral genome. Therefore, it was unclear whether the L2-dependent redistribution of E2 in the BPHE-1 might depend on the presence of the viral genome in the cells.

To determine if the L2-dependent redistribution of E2 observed in the BPHE-1 cells could occur between L2 and the full-length E2 protein, independently of the viral genome, we infected BHK-21 cells (which do not contain the papillomavirus genome) with both the L2-SFV recombinant and an SFV recombinant expressing the full-length E2. Since the RNA for E2 was produced entirely by the SFV RNA-dependent polymerase in the cytoplasm, production of the alternative E2 mRNAs was precluded. As expected, only the 48-kDa form was detected on Western blots of SFV-E2-infected cell extracts (data not shown). As noted above, the L2 distribution in BHK-21 cells was similar to that observed with the BPHE-1 cells (Fig. 5A). When E2 was expressed in BHK-21 cells, in the absence of L2, most of the protein was present in a diffuse nuclear distribution (Fig. 5B). When the cells were coinfected with L2 and E2, the L2 pattern was unaltered but E2 assumed the punctate staining pattern of L2 in the cells that coexpressed the two proteins (Fig. 5C and D). These results indicate that L2-dependent localization of the full-length E2 to PODs is independent of the viral genome and viral gene products other than L2.

L2 does not induce the redistribution of E1.

We also examined the localization of E1, which participates in viral DNA replication and so is presumably expressed in BPHE-1 cells. The immunostaining with an anti-E1 antibody in BPHE-1 cells was weak. This result was expected, since only low levels of E1 expression from steady-state autonomously replicating BPV genomes have been reported (51). No change in the speckled staining pattern was observed after SFV-mediated expression of either capsid protein. Because the intensity of the staining was so low and the parental line of BPHE-1 was not available as a control, no firm conclusions could be drawn from the E1 analysis in these cells.

To overcome these problems, BHK-21 cells were infected with an E1-SFV recombinant, which resulted in clear immunostaining in a speckled nuclear pattern, while uninfected cells were negative (Fig. 6A and B). Coinfection with the L2 and E1 recombinant SFVs resulted in the typical punctate L2 staining pattern (Fig. 6D), but this expression did not alter the E1 pattern in the coinfected cells (Fig. 6C). Therefore, the L2 protein does not directly induce a redistribution of E1. However, these results do not preclude the possibility that E1 localizes to PODs indirectly through its well-documented interaction with E2 (42, 56).