INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Hepatitis delta virus (HDV) is a small RNA virus that is found in nature uniquely associated with human hepatitis B virus (HBV), and coinfection with HDV often increases the severity of HBV-associated liver disease (24, 43). Although HDV RNA replication can proceed independently of HBV (22), HDV does not encode envelope proteins and therefore requires those of its HBV helper for virion assembly, release, and infectivity. The HDV genome is a single-stranded, covalently closed, circular RNA with many similarities to the genomes of plant viroids (4, 46). Unlike viroids, however, it bears a single open reading frame, encoding the hepatitis delta antigen (HDAg). This gene product is expressed during viral infection in two isoforms, small (S-HDAg) and large (L-HDAg). The small antigen is required for viral RNA synthesis (22), which, by analogy with viroids, is thought to proceed via a rolling-circle mechanism (4). Although the identity of the responsible polymerase is unknown, host RNA polymerase II (Pol II) is suspected to be involved (29). The L-HDAg isoform is generated during viral replication by RNA editing of the stop codon of S-HDAg, resulting in the addition of a C-terminal tail of 19 amino acids. The resulting L-HDAg polypeptide has two new functions: (i) it inhibits HDV RNA replication, and (ii) it promotes the envelopment of HDV RNPs by HBV envelope glycoproteins (7, 13, 41).

Several previous studies have examined the subnuclear distribution of HDAg, with various results (2, 9, 29). Delta antigen has variously been described as localized to the nucleolus, the nucleoplasm, or both. In some cases (29), different cells in the same preparation displayed different patterns of localization. Not uncommonly, a punctate pattern of staining is observed, though often in conjunction with more diffuse staining. In an important study, an attempt was made to determine the origin of the punctate staining (2). The authors suggested that transient colocalization of HDAg-containing dots with nuclear depots of the splicing factor SC35 (interchromatinic granule clusters) occurs early in infection; later in infection HDAg staining (while still punctate) was no longer coextensive with SC35-containing structures. The origin of these non-SC35 structures was not determined. A similar study (9) also noted punctate and diffuse accumulation of HDAg (and viral RNA) but did not find any association with SC35 speckles. In this study, L-HDAg occurred preferentially in specific aggregates, while antisera that also recognized S-HDAg reacted both with these aggregates and throughout the nucleoplasm.

Here we have examined the subnuclear distribution of HDV proteins and RNA in more detail. Using HEp-2 cells transiently transfected with 1.1-mer constructs of HDV, a system which supports authentic viral RNA replication and RNA editing, we have generated the full complement of viral replicative intermediates and polypeptides. We have employed antibodies specific for L-HDAg and hybridization probes that distinguish genomic from antigenomic RNA. Moreover, we have attempted to determine the origin of the punctate dots, so often noticed in earlier studies, by costaining for both SC35 and components of the nuclear domain 10 (ND10).

ND10 (also known as PML bodies or PODs) are subnuclear domains containing the putative proto-oncogene product PML (11, 20, 49). This protein, if overexpressed, appears to enhance apoptosis (40), and cells in which PML expression has been eliminated display reduced apoptosis (48). ND10 complexes also contain other proteins, including the apoptosis-enhancing protein Daxx, which appeared to bind to the death domain of the Fas receptor (47, 50). This protein has recently been shown to be important for the formation of ND10 by mediating the deposition of other proteins in PML aggregation sites (16). Both PML and Sp100, another ND10-localized protein, are modified by the small ubiquitin-related modifier 1 (SUMO-1) or sentrin-1 (3, 10, 19, 37, 44). The SUMO-1 modification of PML is essential for the binding of Daxx to PML and, in that sense, for maintenance of ND10 (16).

Our interest in ND10 and HDV infection was sparked by the number and appearance of the HDAg aggregation sites and by the fact that many viruses appear to interact with ND10 during the beginning of their reproductive cycles (30). However, such viruses also encode proteins that alter or disrupt ND10 architecture. The reasons why the initial infecting genomes begin their transcriptional cascade at ND10 are not clear. However, the idea has been advanced that ND10 functions as a depot where excess components can be segregated to remove them from the available pool (30). Here, we show that the large HDV-containing aggregates in infected HEp-2 cells correspond to complexes of L-HDAg and antigenomic RNA, that they contain the repressive ND10-associated protein PML, and that a subpopulation of these HDV complexes indeed are spatially associated with ND10 and partially segregate ND10-associated proteins into virus-induced structures.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Antibodies and cell culture. ND10s were visualized by using various antibodies. Monoclonal antibody (MAb) 5E10 (45) recognizes PML (31). MAb 138 recognizes NDP55 (1). Antibodies against SUMO-1 came from P. Freemont. Isotype-matched MAbs of unrelated specificities were used as controls. Polyclonal rabbit antiserum against PML and Sp100 came from J. Frey (21). Antibodies against Daxx were produced in mice, and one rabbit antiserum against Daxx was bought from Santa Cruz (Palo Alto, Calif.). RNA Pol II was localized using antibodies provided by S. L. Warren (6). Antibodies against a presently uncharacterized ND10 protein were found in rabbit serum originally raised against delta-interacting protein A. The L-HDAg-specific antibodies were produced in guinea pigs that had been immunized with a peptide conjugated to keyhole limpet hemocyanin WDILFPADPPFSPQSCRPQ (Animal Pharm Services), which corresponds to amino acids 196 to 214 of L-HDAg. For detection of both L- and S-HDAg, rabbit antibodies raised against recombinant S-HDAg (5) were used. HEp-2 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics. All cells were grown at 37°C in a humidified atmosphere containing 4% CO₂. For immunohistochemical analysis, cells were grown on round coverslips in 24-well plates (Corning Glass, Inc.) until approximately 80% confluence before fixation.

Immunolocalization and in situ hybridization. One day after being plated, the HEp-2 cells were transfected with plasmid pDL538, which produced HDV genomic RNAs that served as templates to initiate HDV replication. (S-HDAg is transiently expressed from this same plasmid using a fortuitous promoter within the plasmid sequences.) pDL538 is identical to pDL542 (25) except that it lacks the 2-nucleotide deletion present in pDL542. In some experiments, another plasmid, pDL456, was used for transfection, which produced antigenomic instead of genomic RNAs. PDL456 is identical to pDL481 (26) except that it carries the wild-type HDV sequence without the 2-nucleotide deletion of pDL481. All transfections were performed using the DOSPER reagent (Boehringer Mannheim) according to the manufacturer's instructions. The cells were fixed at different intervals after transfection and assayed by immunofluorescence using different antibodies to localize various proteins and/or by in situ hybridization to localize the HDV genomic and antigenomic RNA molecules. When in situ hybridization was combined with immunocytochemical procedures, we refixed the cells after incubation with the first antibodies before denaturation of the partially double-stranded HDV RNAs. The cells were fixed at room temperature (RT) for 15 min using freshly prepared 1% paraformaldehyde (PFA) in phosphate-buffered saline (PBS), washed with PBS, and permeabilized for 20 min on ice with 0.2% (vol/vol) Triton X-100 (Sigma Chemical Co.) in PBS. Antigen localization was determined after incubation of the permeabilized cells with rabbit antiserum or with MAb diluted in PBS for 1 h at room temperature. Fluorescein- or Texas red-labeled anti-mouse or anti-rabbit immunoglobulins (Vector Laboratories) were complexed with primary antibodies for 30 min. The cells were then stained for DNA with 0.5 µg of bis-benzimide (Hoechst 33258; Sigma)/ml in PBS and mounted with Fluoromount G (Fisher Scientific).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Distribution of L-HDAg in the nucleus. To investigate the nuclear distribution of HDV replication products, we transfected HEp-2 cells with plasmid pDL538, in which genomic HDV RNA is expressed from a simian virus 40 promoter in the vector. Transcription of the HDAg coding sequences within the plasmid also gives rise to S-HDAg, allowing true viral RNA replication to occur. Antigenomic RNA can thus be made from the genomic RNA template. Further rounds of replication amplify both genomic and antigenomic RNA copies from their respective RNA templates. As replication proceeds, additional S-HDAg can be expressed from small mRNAs of antigenomic polarity. RNA editing of these transcripts results in the generation of L-HDAg, which acts to down-regulate replication. Thus, cells transfected with pDL538 can engender the entire HDV replication cycle.

View larger version (32K): [in this window] [in a new window]   FIG. 1.   Immunohistochemistry and in situ hybridization 4 days after HDV transfection of HEp-2 cells. The upper corners of each image show the color used for the respective protein or RNA: green, fluorescein; red, Texas red; blue, Hoechst stain for DNA. HDA denotes L-HDAg. (A) Single nucleus shown double labeled for DNA and L-HDAg. L-HDAg (top) is situated over an area devoid of DNA (see the same nucleus at the bottom). (B) Two nuclei double labeled for L-HDAg and SC35, showing that L-HDAg occupies a different space than SC35. (C) Mitotic cell labeled for SC35 and L-HDAg. SC35 is dispersed but excluded from chromosomes in the metaphase plate (vertical empty space in the cell). The L-HDAg aggregates are retained through mitosis. (D) Single nucleus double labeled for L-HDAg and PML, showing that many L-HDAg accumulations are found beside ND10. (E) Double-labeled single nucleus where Bcl6 aggregates do not have substantial association with ND10. (F) Double-labeled single nucleus showing that PML and L-HDAg occupy the rims of apparently hollow structures. (G) Single nucleus producing L-HDAg. L-HDAg and Sp100 labeling indicates that Sp100 is located inside the L-HDAg spheres. (H) Low magnification of cluster of cells producing L-HDAg, demonstrating that transfected cells formed a clone over 4 days and that Daxx locates beside L-HDAg. (I) Higher magnification of the same preparation as in panel G, showing two nuclei. In the lower one, Daxx is located in the inside of L-HDAg spheres and in small dots on their outsides. (J) SUMO-1 and L-HDAg labeling showing that SUMO-1 is present throughout the L-HDAg spheres. (K) In situ hybridization showing that the dominant distribution of genomic RNA is diffuse and not associated with ND10 4 days after transfection (lower cell; the upper cell is untransfected). (L) A small number of cells have some genomic RNA aggregates associated with ND10, as indicated by Sp100, but genomic RNA does not colocalize with ND10. (M) In situ hybridization to label antigenomic RNA and L-HDAg. The components colocalize. (N) Staining by in situ hybridization of antigenomic RNA and immunolabeling of Sp100. Sp100 is found in the center of the antigenomic RNA sphere. (O) Location of an as-yet-uncharacterized ND10-associated protein (ND10P) inside the antigenomic RNA sphere 4 days after transfection. (P) In situ hybridization to label antigenomic RNA transcribed directly from the transfected plasmid and L-HDAg. Antigenomic RNA also associates with L-HDAg if transcribed from DNA.

Spatial localization of L-HDAg aggregates relative to ND10. Because L-HDAg appeared initially as precise aggregates in the nucleus, we compared the positions of ND10-associated proteins relative to L-HDAg aggregates. We used antibodies to the following known components of ND10: PML, the PML-interacting protein Daxx, Sp100, SUMO-1 (which can covalently modify some ND10 proteins), and several other as yet molecularly uncharacterized ND10-associated proteins. In double-labeling experiments with anti-L-HDAg and anti-PML antibodies, we found that L-HDAg aggregates are often localized adjacent to ND10, as indicated by staining for PML (Fig. 1D). Although 96% of transfected cells displayed some of this spatial association of HDAg aggregates and ND10, within any given transfected cell, not all of the L-HDAg aggregates were directly associated with PML. Quantitation revealed that, on average, approximately 55% of all recognizable L-HDAg aggregations were associated with ND10. Interestingly, cells with L-HDAg expression generally exhibited a lower number of ND10s, which in turn were typically smaller, whether or not they were associated with the L-HDAg aggregates.

Change in ND10-associated protein distribution by HDV infection. Several viruses have been observed to modify the distribution of ND10-associated proteins differentially (12, 30, 32, 33, 39). We observed that L-HDAg and PML often colocalized when the HDAg aggregate became large and appeared hollow (Fig. 1F). It appears as if PML was repositioned, resulting in smaller and fewer ND10s. Therefore, we tested whether another ND10-associated protein, Sp100, was changed in its position relative to PML. Double labeling of Sp100 and L-HDAg showed an Sp100 distribution quite different from that of PML in these spheres. L-HDAg surrounded Sp100 (Fig. 1G). Although the projection of a small Sp100 site with the hollow HDAg sphere (but located outside of the sphere) could result in such an image, this seems unlikely, as the resolution in the z axis is better than 0.7 µm and the spheres approach 3 µm (resolution in the x-y dimension is slightly better than 0.3 µm). Overlaps of spatially separate volumes would generate images, as seen in Fig. 1D, upper left. The resolution is, however, not high enough to exclude a substantial overlap between the apparent rim and what is described as the center of such a sphere. The same central distribution was found with two additional, as-yet-uncharacterized ND10-associated proteins (data not shown).

Antigenomic HDV RNA colocalizes with L-HDAg. We next investigated where the different RNA replication products of HDV would localize in the nucleus relative to ND10 and the L-HDAg sites by examining transfected cells for viral replicative intermediates of each polarity by in situ hybridization. To obtain a strong hybridization signal, the partially double-stranded HDV genomic and antigenomic RNAs needed to be denatured before hybridization. This created potential problems with the probes hybridizing to the input DNA. Therefore, we digested all DNA with DNase I before hybridization. Control cells and cells infected for 1 day were all negative both for viral RNA (by in situ hybridization) and for L-HDAg (as judged by immunofluorescence). However, 3 days after transfection, we found many cell clusters that showed punctate labeling for viral RNA (see below).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Although the architecture of the mammalian nucleus is incompletely understood, it is clearly both complex and dynamic. Numerous different nuclear substructures or domains have been identified by morphological, biochemical, and immunocytochemical approaches. ND10s represent an important and intensively studied example of such substructures. While their function(s) remains controversial, it is known that they are dynamic structures that change considerably during viral infection. In the case of DNA viruses, ND10s represent early sites of viral transcription and replication (17, 18, 30, 34). During the replication cycle of both adenoviruses and herpesviruses, ND10s undergo substantial reorganization. To our knowledge, nothing is known about the role of ND10s in the replication of RNA viruses and the effects of RNA virus replication on these structures, although PML overexpression reduces replication of vesicular stomatitis virus and influenza A virus (8). Here we have examined the distribution of HDV RNAs and proteins in infected cells, as well as the effects of viral replication on ND10 structure. We find that in epithelial cells in which HDV replication is ongoing, L-HDAg and antigenomic RNA are colocalized in novel and discrete nuclear substructures, while genomic RNA strands and a substantial amount of S-HDAg are distributed more broadly in the nucleus. Because we were able to observe a diffuse staining pattern for S-HDAg even soon after transfection and in cells with low staining intensities for total HDAg, we believe that the failure to detect L-HDAg outside of its aggregations is not due to a lower concentration of L-HDAg than of S-HDAg. An identical dotlike distribution pattern of L-HDAg has been observed by us and others in the human hepatoma cell line Huh7, indicating that these structures do not represent a storage artifact of L-HDAg that is only present in HEp-2 cells. Roughly half of these L-HDAg antigenomic RNA aggregations appear to associate with components of ND10. In such complexes, the organization of the constituent ND10 components appears to have been altered in comparison to that of normal ND10s in uninfected cells.

These results can be instructively compared to those of earlier localization studies of HDV antigens and RNAs. As in earlier studies, we observe that HDAg can be found diffuse and in aggregates. In HEp-2 cells, it is the large isoform of HDAg that is predominantly localized to these complexes, though we cannot exclude the presence of a small fraction of the S-HDAg isoform there. Clearly, a substantial fraction of S-HDAg (the dominant isoform) must be more broadly distributed. This result is similar to what was observed by Cunha et al. (9) in a stabile transfected Huh7 cell line bearing HDV replicative intermediates, though the localization of L-HDAg is more pronounced in HEp-2. Like Cunha et al. (9), we do not observe clear colocalization of HDV antigens with SC35 speckles, as initially reported by Bichko and Taylor (2). Rather, expanding L-HDAg sites appeared to displace SC35 domains, resulting in an excluded space outside the SC35 domain. However, Bichko and Taylor (2) also reported that the HDAg-SC35 association was transient and was replaced later by punctate HDAg dots of unknown provenance. We think it highly likely that the L-HDAg aggregates we describe here correspond to those non-SC35 structures.

Our studies establish that many (but not all) of these L-HDAg-containing structures have some association with proteins known to compose ND10. It appears as if smaller L-HDAg aggregates are predominantly located beside ND10 and ND10-associated proteins are recruited from there into the expanding HDAg antigenomic-RNA-containing spheres. This change in distribution differs significantly from the case observed in DNA virus infection, where specific proteins aid in the destruction of ND10, except for Ad5, where PML is segregated into separate tracks and the other ND10-associated proteins are recruited into the replication domain (30). But, as in DNA virus infection, there did appear to be significant changes in ND10s resulting from the presence of HDV. First, there was a reduction in size of ND10s posttransfection, concomitant with an increase in the size of the L-HDAg sites. In parallel there was a relocalization of ND10-associated proteins from ND10s into L-HDAg spheres. This relocation involved all the tested ND10-associated proteins, including the newly defined Daxx (16). At smaller (presumably younger) L-HDAg sites, the ND10 proteins appeared to remain outside the L-HDAg aggregates. With increasing size of the L-HDAg accumulation, ND10-associated proteins appear to commingle with the viral antigen, but when these sites enlarge further, they develop into more structured spheres in which the distribution of ND10-associated proteins diverges visibly, with PML being found in the rim and the other proteins in the center of the sphere. Some overlap may exist, both real and due to the resolution of the images, which is not always sufficient to separate the projection of the spheres' rims from their centers.

We do not know the basis for the segregation of ND10 proteins in this setting; perhaps L-HDAg displaces the Daxx interaction with SUMO-1-modified PML, releasing all other ND10-associated proteins dependent on the PML-Daxx interaction for their segregation. Only SUMO-1 seems to straddle both the rim and the inside of the sphere, which is explainable by the fact that two known ND10-associated proteins, PML and Sp100, can both be modified by SUMO-1.

We often observed clusters of HDAg-positive cells, suggesting that some cells may have undergone several divisions following infection. Also, the RNA and L-HDAg production was lower than expected from the appearance of other transfected proteins within the same time frame. The cells were not congested, which may have allowed continuous cell proliferation. Thus, HDAg appears to be produced in a highly controlled way, avoiding overexpression. This may also explain why we did not detect large amounts of HDAg-mRNA in the nuclei of transfected cells. In another difference from ND10 in uninfected cells, L-HDAg aggregates did not appear to undergo total disassembly during mitosis (most ND10s disperse beginning with nuclear envelope breakdown). Normally, large aggregates are excluded during nuclear envelope formation. Here it is not clear whether the HDAg aggregates are reintroduced into the nucleus. Their potential reenvelopement would, however, explain the fact that half of the HDAg aggregates were not found in association with newly forming ND10s, assuming that their initial aggregation took place there. The significance of the spatial closeness but obvious separateness of many small HDAg aggregates and ND10s is not apparent but is underscored by the acquisition of ND10-associated proteins in HDAg spheres.

Another novel and surprising finding was the preferential association of L-HDAg with antigenomic RNA. In vitro, recombinant L-HDAg can bind to synthetic HDV transcripts of both genomic and antigenomic polarities (7, 27, 28), and one earlier in situ hybridization analysis of Huh7 cells did not distinguish between the two RNA polarities (9). The study was done with cells transformed with HDV genomes, a circumstance that we have found does not usually permit authentic HDV replication (D. Wang, J. Pearlberg, and D. Ganem, unpublished data); results with such lines, therefore, may not be representative of more typical HDV replication events.

The striking colocalization of L-HDAg and antigenomic RNA suggests a potential function for the complex. As noted earlier, L-HDAg can inhibit viral RNA synthesis as it accumulates during infection; since it is also required for packaging of the viral genome, its accumulation signals a switch from replication to envelopment. We suspect that the L-HDAg complexes we observe may be the sites of the L-HDAg-mediated repression of genomic RNA production from antigenomic templates. Consistent with this, we detect no staining for RNA Pol IIthe enzyme presumed responsible for HDV RNA replicationin these complexes (see also reference 9). If this is the true role of the complexes, then perhaps one mechanism by which L-HDAg inhibits replication is to target host proteins with the capacity for repression of Pol II-based transcription to the viral RNA. In this connection, we note that PML and Daxx were reported to be transcriptional repressors (14, 36). All of the larger spheres are thus associated with ND10-associated proteins. The smaller and presumably earlier stages of antigenomic RNA accumulations are partly associated with ND10, but the ND10-associated proteins are not part of the antigenomic aggregates. Also, about half of these aggregates are not associated with ND10. They may have originally started at ND10 but may have lost this connection through mitotic events, after which they are not repositioned at such newly forming sites.

While we cannot exclude potential functional roles for the complex, it seems unlikely that the observed L-HDAg-RNA complexes play a direct role in nuclear export of RNPs or in their envelopment. Although L-HDAg is required for virion envelopment and release, this activity should result in the production of genomic RNAs bound to L-HDAg and S-HDAg. In fact, on the surface, the failure to find L-HDAg associated with genomic RNA appears paradoxical. Since L-HDAg is responsible for virion envelopment, and since virions are composed principally of genomic RNA, one might anticipate finding complexes of L-HDAg and genomic HDV RNA. However, we note that our morphological studies identify only the most abundant accumulations of L-HDAg. We think it likely that the much lower levels of L-HDAg likely to be found on genomic RNA do not allow visualization by immunohistochemistry.

The previously unsuspected colocalization of L-HDAg with antigenomic RNA and PML, as well as the segregation of other ND10-associated proteins, raises new possibilities for the regulation of both HDV replication and ND10 function. Analysis of the molecular interactions of L-HDAg with ND10 componentsand of the latter with viral RNPsshould shed new light on both processes.