|Home | About | Journals | Submit | Contact Us | Français|
Hepatitis delta virus (HDV) encodes one protein, hepatitis delta antigen (δAg), a 195-amino-acid RNA binding protein essential for the accumulation of HDV RNA-directed RNA transcripts. It has been accepted that δAg localizes predominantly to the nucleolus in the absence of HDV genome replication while in the presence of replication, δAg facilitates HDV RNA transport to the nucleoplasm and helps redirect host RNA polymerase II (Pol II) to achieve transcription and accumulation of processed HDV RNA species. This study used immunostaining and confocal microscopy to evaluate factors controlling the localization of δAg in the presence and absence of replicating and nonreplicating HDV RNAs. When δAg was expressed in the absence of full-length HDV RNAs, it colocalized with nucleolin, a predominant nucleolar protein. With time, or more quickly after induced cell stress, there was a redistribution of both δAg and nucleolin to the nucleoplasm. Following expression of nonreplicating HDV RNAs, δAg moved to the nucleoplasm, but nucleolin was unchanged. When δAg was expressed along with replicating HDV RNA, it was found predominantly in the nucleoplasm along with Pol II. This localization was insensitive to inhibitors of HDV replication, suggesting that the majority of δAg in the nucleoplasm reflects ribonucleoprotein accumulation rather than ongoing transcription. An additional approach was to reevaluate several forms of δAg altered at specific locations considered to be essential for protein function. These studies provide evidence that δAg does not interact directly with either Pol II or nucleolin and that forms of δAg which support replication are also capable of prior nucleolar transit.
Hepatitis delta virus (HDV) is a subviral agent with a single-stranded circular RNA genome. As reviewed recently (6, 14, 37) this 1,679-nucleotide RNA is replicated by RNA-directed RNA synthesis dependent upon redirection of the host RNA polymerase II (Pol II). Replication involves the production of the antigenome, an exact complement of the genome. The small delta antigen (δAg) is translated from a third RNA, one with the same antigenomic polarity but with a linear rather than circular conformation, of less than full-length, and possessing a 5′ cap and a 3′ poly(A). δAg is a 195-amino-acid-long RNA binding protein that is essential in one or more ways for the accumulation of HDV RNA-directed RNA transcripts. There is a second form of δAg that is 19 amino acids longer at the C terminus; this species arises due to specific RNA editing that occurs on the antigenomic RNA during replication. Unlike the small form, the large does not support HDV genome replication and is a dominant-negative inhibitor of the replication driven by the small form. However, large δAg does have an essential function in that in the presence of HBV envelope proteins, it facilitates the assembly of HDV genomic RNA into new virus particles.
Our previous immunostaining studies have shown that in the absence of HDV genome replication, small δAg localizes predominantly to the nucleolus, with some staining in the nucleoplasm (4). It has been reported that δAg moves to the nucleus because of a bipartite nuclear localization signal (NLS) (21); however, a more recent study shows that a single short sequence (positions 66 to 75) is sufficient to direct a marker protein to the nucleus, where it localizes to the nucleoplasm (2). Nucleolar localization of δAg has been attributed to a nucleolar localization signal (24). It has been suggested that δAg binds directly to nucleolin (24), also known as C23, and to nucleophosmin, also known as B23 (17). The present study includes examination of the nature and significance of the δAg localization to the nucleolus. Other investigators have noted that for several viruses, specific and essential viral proteins, typically RNA-binding proteins, also locate to the nucleolus, and in some cases such localization only occurs early in infection, leading to the speculation that a “nucleolar transit” may be an essential step in genome replication (12, 15). Consistent with but not proof of such a transit for δAg is that during HDV genome replication, the majority of δAg is no longer found in the nucleolus but in the nucleoplasm (4). Located within the nucleoplasm is the majority of the host RNA Pol II, and so it has been tempting to suggest that colocalization of δAg might be an indication of HDV RNA-directed transcription complexes (8). However, the present studies indicate that even though the majority of the δAg is in the nucleoplasm, it is not in association with active HDV RNA-directed transcription by Pol II but rather, with the accumulation of completed transcripts. Furthermore, even nonreplicating HDV RNAs can move δAg from the nucleolus to the nucleoplasm.
Along with other investigators we have noted that during HDV replication, there arise mutated forms of δAg. There is the large δAg that arises via editing. However, there are other forms, some of which might support infection or coexist in the presence of forms that do (3, 4, 13). Such mutated forms present a spectrum of different intracellular localizations; for example, they may be spread throughout the whole cell or located over the endoplasmic reticulum or associated with the splicing factor SC35 at speckle sites in the nucleoplasm. Therefore, for most of the present study, we focused on expressing the essential small δAg, without and with replicating or nonreplicating HDV RNAs, under conditions designed to prevent altered forms of the protein from arising. Then, with immunostaining procedures we examined the intracellular localizations of δAg relative to specific cellular components, such as nucleoplasmic Pol II and SC35 or nucleolar nucleolin. Our findings have implications for the hypothesis of prior nucleolar transit of δAg and for the significance of δAg in the nucleoplasm when HDV RNA replication occurs.
Two inducible cell culture systems were as previously described (7). Briefly, we first established from TRex cells (Invitrogen) a clone, 293-δAg, expressing a single copy of δAg cDNA under tetracycline (TET) control. Next, HDV genome replication was initiated in these cells by transfection of HDV RNA that has a frameshift mutation and does not express δAg by itself. A clone derived from a single cell was identified by its high level of HDV replication and was designated as 293-HDV cells. Both 293-δAg and 293-HDV cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and selective antibiotics, blasticidin and hygromycin (Invitrogen). Also used were 293T and Huh7 cells (30).
Vectors expressing the forms of δAg and HDV were as previously described (22, 23, 28). Transfection of 293T and Huh7 cells was via Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.
Indirect immunofluorescence was performed as previously described (8). After paraformaldehyde fixation, we used the following primary antibodies: polyclonal rabbit anti-δAg and mouse monoclonal antibodies specific for nucleolin, Pol II (4H8), and splicing factor SC35. The mouse antibodies were obtained from Abcam. As secondary antibodies we typically used Alexa 488 chicken anti-mouse and Alexa 594 goat anti-rabbit (Invitrogen). Cells were stained with DAPI (4′,6′-diamidino-2-phenylindole) to visualize the nuclei and mounted with antifade mounting solution (Invitrogen). Confocal images were obtained using either a Zeiss 510 laser scanning microscope or a Nikon Eclipse TE2000-E. Optical sections of 0.1 μm were collected. In some cases Z-stacks at 0.25 μm were collected, and the intracellular distribution of stained proteins was determined using MetaMorph software.
As described previously (7) and in Materials and Methods, both the 293-δAg and 293-HDV cell lines contain a TET-inducible cDNA for the small δAg. In addition, the 293-HDV cells contain copies of HDV RNA so that TET induction leads to increased transcription and accumulation of HDV RNAs. At 16 and 40 h after induction, the immunostaining patterns were determined as shown in Fig. Fig.11.
For 293-HDV cells the δAg pattern was the same at both 16 and 40 h. It was predominantly nucleoplasmic, colocalizing with host RNA Pol II. We previously showed that HDV RNA accumulation reaches maximal values by 40 h (7). Thus, we consider that for the majority of the δAg in these cells, the nucleoplasmic distribution reflects association with the accumulation of processed HDV RNA species rather than with the process of RNA-directed transcription.
At 16 h the predominant distribution of δAg in 293-δAg cells was with host nucleolin, a major nucleolar protein, rather than with Pol II, a nucleoplasmic protein. A minority of cells gave staining in both nucleoli and nucleoplasm or in nucleoli only. After 40 h the predominant staining pattern was in nucleoplasm only. Quantitation of the three staining patterns is presented in Table Table11.
For 293-δAg cells at 40 h the nucleoplasmic distribution of δAg seemed similar to that obtained for 293-HDV cells, and yet in this case it had nothing to do with HDV RNA-directed RNA transcription. We noted that the nucleolin distribution in 293-δAg cells also changed with time, from being nucleolar to more nucleoplasmic. This change also occurred in 293T cells that lack the δAg (data not shown). Thus, we considered that for both the δAg and the nucleolin, the change in localization might actually be associated with the time since the cells were seeded. As confirmation of this interpretation, we found that when induced cells were reseeded at subconfluent density and observed within 24 h, the majority of both δAg and nucleolin localized to the nucleolus (data not shown). In an earlier study of δAg expression in Huh7 cells at 6 days after transfection, we observed δAg to be predominantly nucleolar; however, such cells had been reseeded at subconfluent density within 24 h of observation (4).
Upon further study, we found two situations that quickly and reversibly changed the intracellular distribution of δAg and nucleolin in 293-δAg cells. At 16 h after induction cells were treated for 1 h with either actinomycin or 5,6-dichloro-1-beta-d-ribofuranosylbenzimidazole (DRB). A low concentration of actinomycin (10 ng/ml) is known for its ability to specifically block transcription by Pol I of rRNA precursors (31). DRB is a nucleoside analog known for several activities in animal cells. It inhibits kinases including casein kinase II and in this manner can interfere with Pol II transcription (44). It is also known to be a cause of nucleolar stress (18). Nucleolin is known to be hyperphosphorylated by kinases including casein kinase II (26). δAg is also reported to be phosphorylated by this kinase (42).
As shown in Fig. 2B and D, treatments for 1 h with actinomycin or DRB, respectively, caused the majority of the δAg to be detected in the nucleoplasm. The same treatments also caused much of the host nucleolin to be localized in the nucleoplasm. Furthermore, if the treated cells were then returned to normal growth conditions for 2 h, within each cell a significant fraction of both δAg and nucleolin were back in the nucleolus, as shown in Fig. 2C and E, respectively. A similar effect for DRB has been reported for a fusion of large δAg with green fluorescent protein (27, 35).
The studies shown in Fig. Fig.22 support the interpretation that for induced 293-δAg cells, δAg, similar to what is known for nucleolin, was in the nucleolus because of binding to rRNA precursors. Also, the studies shown in Fig. Fig.1,1, with the time-dependent movement of δAg and nucleolin to the nucleoplasm, are consistent with the maturation of the rRNA precursors and the release of δAg from the nucleoli. Nevertheless, while these interpretations might explain why the δAg could leave the nucleolus, they do not explain why the δAg (or nucleolin) was then relocated to the nucleoplasm rather than to another cellular compartment.
The above studies suggested that δAg might bind to rRNA precursors in the nucleolus. As a test of this interpretation, we examined the consequences of also expressing nonreplicating HDV RNAs since we might expect that δAg would prefer to bind to the latter. Huh7 cells, a human liver cell line, were transfected with a plasmid to express δAg (pDL444), without or with a plasmid to express an excess of full-length HDV RNA. The latter, plasmid pDL482, as previously described, was a chimera of genomic and antigenomic HDV RNA sequences and was incapable of replication even when provided with δAg (23). After 24 h cells were examined for the distribution of δAg and nucleolin.
For cells expressing δAg we detected only nucleolar localization (Fig. (Fig.3A).3A). However, for about half of the cells coexpressing the HDV RNA, we detected two additional patterns: localization in nucleoli and nucleoplasm or in the nucleoplasm only. Table Table22 gives the quantitation of this and additional experiments where we tested the consequences of expressing other RNAs. Another HDV RNA, lacking sequences from one end of the genomic rod-like structure, also caused redistribution of δAg.
Thus, nonreplicating HDV RNAs were able to cause movement of δAg from the nucleoli to the nucleoplasm. The results support the interpretation that, given the choice, δAg will accumulate on HDV RNAs even if they are nonreplicating rather than on rRNA precursors.
In contrast to 293-δAg cells, when induced 293-HDV cells were treated for 1 h with actinomycin or DRB, there was no detectable change in the δAg distribution even though there was movement of nucleolin from the nucleolus (data not shown). Furthermore, when the cells were briefly treated with amanitin (45 min at 5 μg/ml), an inhibitor of host RNA Pol II and known to be able to inhibit HDV RNA accumulation (9, 29), there was again no change in the distributions of δAg or of Pol II (Fig. 4A and B). Such findings provide more support for the interpretation that in induced 293-HDV cells, the distribution of the majority of δAg has nothing to do with HDV RNA transcription although it may have something to do with the accumulation of processed HDV RNAs. However this does not exclude the possibility that even after disruption of transcription, the intracellular localization of δAg and RNA remains nucleoplasmic.
It has been reported that δAg in vitro interacts with an essential transcription factor, pTEFb, a complex of CDK2 and cyclin T (40). Also, it is reported that in animal cells much of the Pol II can exist bound to specific templates but that in a paused state it is associated with a depletion of active factor pTEFb and, further, that this depletion is caused by the association of pTEFb to 7SK RNA in the nucleolus (43). One way to disrupt the nucleolar complex and release paused Pol II is by treating cells with HMBA (10, 32). We applied such a treatment during TET induction of 293-HDV cells. As shown in Fig. Fig.4C,4C, this caused no significant change in the intranuclear distribution of δAg or Pol II, and it had no effect on the extent of accumulation of HDV RNA (data not shown). This supports the interpretation that δAg in the nucleoplasm was not in a paused configuration with Pol II.
Figure Figure55 is a representation of features that have been described for the unmodified δAg, with alignment on the 195-amino-acid species used in our studies (19). At position 12 to 60 is a domain that was noted to have a potential alpha-helical structure (45). Initial studies to determine why δAg located to the nucleus defined a bipartite nuclear localization signal. Subsequent studies have shown that just one such region at position 66 to 75 (indicated as an NLS) is sufficient (2). Other studies have determined two regions, at positions 97 to 107 and 136 to 146, that are rich in positively charged amino acids, which function as a bipartite RNA-binding domain (RBD) (20). Another study has reported that within the 12 amino acids at the N terminus is yet another RBD (39).
Also indicated in Fig. Fig.55 are five altered forms of δAg that we have previously described (23, 28). These are representative of changes in the features of the unmodified wild-type δAg. These proteins were transiently expressed by transfection of 293 cells, and then the intracellular distributions were examined by immunostaining.
Figure Figure6A6A shows the results for two forms of δAg that were altered in the coiled-coil domain (CCD). The mutant with a deletion of residues 19 to 31 (Δ19-31) fails to dimerize (23, 28). The amino acid exchange mutant, W50A, is predicted to be important in the CCD structure, and yet experimentally dimerization can still be detected (28). Both altered forms fail to support HDV genome replication (28) and, as shown in Fig. Fig.6A,6A, localized to the nucleoplasm rather than to the nucleolus. Similar results were obtained with W20A that, like W50A, is predicted to be important for the CCD structure and yet does not prevent dimerization (28).
Figure Figure6B6B shows the results for a form, Δ89-145, altered by deletion that includes the two RBDs. A form with a larger deletion, Δ69-146, now including the NLS, also localized to the nucleoplasm. Neither of these species is capable of supporting HDV RNA accumulation (23). Since both species lack the two RBDs, it may be unlikely that they are binding to RNA species in the nucleoplasm. Some of the nucleoplasmic δAg seemed to colocalize to speckles containing the splicing factor SC35. Such speckles are considered to contain Pol II and are adjacent to paraspeckles, putative sites for the processing of Pol II RNA-directed RNA transcripts (5, 33). Shih et al. have previously shown that a green fluorescent protein fusion to the N-terminal 88 amino acids of small δAg primarily localizes to SC35 speckles (36).
Finally in Fig. Fig.6B6B are shown results obtained with a form of δAg, Δ69-74, that has a deletion of most of the NLS (23). This protein was localized mainly to the cytoplasm. A small amount was detected in the nucleus, primarily colocalized with nucleoli. We were initially puzzled to find that this protein has some ability to support HDV transcription and accumulation of HDV RNA transcripts. Relative to the HDV RNA transcript accumulation achieved with unmodified δAg, such accumulation was about 20% (data not shown). Therefore, we used analysis of a series of Z-sections obtained by confocal microscopy to determine the fraction of the protein that was nuclear. It was determined to be about 30%. Thus, relative to unmodified δAg, the altered protein lacking an NLS led to about equal amounts of δAg nuclear accumulation and of HDV RNA accumulation.
This report focuses on the use of immunostaining procedures to study the associations of δAg in the presence and absence of HDV RNAs. The advantage of this approach is that if the cells are fixed prior to staining, one can be confident that observed associations are not artifacts of rearrangement, such as can occur during cell fractionation procedures. This is a concern in our previous studies of δAg that made use of fractionation followed by rate zonal sedimentation, immunoprecipitation, and gel electrophoresis under nondenaturing conditions (8, 11). Even when fixation is followed by fractionation, there are concerns about the interpretability of the complexes detected (8). However, even for studies using fixation followed by immunostaining, there is still the limitation that a colocalization of two components supports but is not sufficient to prove an interaction, whether direct or indirect interactions.
Most studies reported here made use of expression in 293 cells, but many aspects have been confirmed in Huh7 cells. Also, some of the results with the small δAg have been reproduced with large δAg. A difference was that when large δAg was relocalized to the nucleoplasm, it showed a greater tendency than small δAg to localize to SC35 speckles (data not shown).
There are three major questions addressed by this study. First, what are the nature and significance of δAg in the nucleolus? δAg, like nucleolin, is an RNA-binding protein, and we conclude that both proteins colocalize because they bind to rRNA precursors (15); immunoprecipitation of δAg with nucleolin has been reported, but such studies did not exclude the possibility that the interaction was mediated by rRNA (24). Since δAg is nucleolar only in the absence of HDV RNA transcription and/or accumulation, one might argue that nucleolar localization is irrelevant to replication. An alternative interpretation is that δAg goes to the nucleolus before it is utilized to facilitate the accumulation of processed HDV RNA transcripts. If such a nucleolar transit does occur, this raises the question of whether such a transit is somehow required for the viral life cycle. It can be said that the only forms of δAg that support replication are ones which, expressed in the absence of HDV RNAs, will accumulate in the nucleolus. At the same time, altered forms, such as large δAg that does not support replication, will also go to the nucleolus. As suggested in the introduction, we know that the RNA binding proteins of many viruses, both RNA and DNA viruses, can localize to the nucleolus (15, 27). Furthermore, some studies consider nucleolar transit to be of relevance. For example, in the case of orthomyxoviruses there are studies suggesting that prior nucleolar localization is required for the nucleocapsid protein early in replication, which is followed by a combination of nucleoplasmic and nucleolar localizations (12, 34).
Second, what are the nature and significance of δAg found in the nucleoplasm? Several situations have been found in which δAg localized to the nucleoplasm. Not just replicating HDV RNA (Fig. (Fig.1)1) but also nonreplicating HDV RNA (Fig. (Fig.3)3) can cause nucleoplasmic accumulation of δAg. Also, nucleoplasmic accumulation can occur in the absence of detectable HDV RNA species. For example, at 40 h after induction δAg is primarily localized to the nucleoplasm (Fig. (Fig.1).1). Certain cell stress conditions lead to nucleoplasmic localization (Fig. 2B and D). Also, several altered forms of δAg localize to the nucleoplasm (Fig. 6A and B). In each of these situations we do not know what the δAg is binding to in the nucleoplasm. In some cases we can detect a component that colocalized with SC35 (Fig. (Fig.6B).6B). This could reflect indirect rather than direct binding, but the localization apparently does not need δAg to possess either a dimerization or an RBD.
Third, does δAg bind, whether directly or indirectly, to nucleoplasmic Pol II? Some form of binding, direct or indirect, is plausible since it has been shown that Pol II is essential for at least some of the HDV RNA-directed transcription (8). There is a report that δAg can interact in vitro with one or two subunits of host Pol II (41) while a more recent study reconstituted Pol II transcription in the absence of δAg (1). Be that as it may, our previous studies using immunoprecipitation indicate that for the majority of δAg expressed within cells in the presence or absence of HDV RNA, there is only minimal association with Pol II (8). Furthermore, the immunostaining studies show that in the absence of HDV RNAs, the δAg is largely nucleolar while Pol II is nucleoplasmic (Fig. (Fig.1);1); thus, δAg is not binding to Pol II and is somehow binding to nucleolar components, which we believe are probably rRNA precursors. Contrary interpretations are that δAg in the nucleolus is binding to nucleolin (24) and B23 (17), that it has fewer posttranslational modifications than nucleoplasmic δAg (38), that δAg modified to contain a nucleolar localization signal supports the synthesis of antigenomic RNA but not genomic RNA (16), and that Pol I in the nucleolus carries out the transcription of antigenomic RNA (25).
In the presence of HDV RNAs the immunostaining results show that δAg and Pol II are in the same nucleoplasmic location, and it has been tempting to assert that they must be associated and participating in RNA-directed RNA synthesis (8). In contrast, after the present studies we would be more cautious. First, at 40 h after induced replication, when accumulation of processed HDV RNAs has reached a maximum value and transcription may have ceased (7), the same colocalization is observed (Fig. (Fig.1).1). Furthermore, when HDV transcription is inhibited with amanitin, the colocalization remains (Fig. (Fig.4).4). Therefore, we suggest that the majority of δAg that colocalizes with Pol II in the nucleoplasm is not currently involved with active RNA-directed transcription. In cells that have undergone HDV transcription, we have previously used immunoprecipitation and found at least 16% of the HDV RNAs in association with δAg (8). Thus, we expect that the majority of δAg in the nucleoplasm is in association with HDV RNAs in a postreplicative state. However, we do not know why such ribonucleoprotein complexes remain largely in the nucleoplasm. In addition, we have to remember that δAg can have a similar nucleoplasmic localization in the presence of HDV RNAs that cannot replicate (Fig. (Fig.3)3) or in the total absence of full-length HDV RNAs (Fig. (Fig.11 and and2),2), and so also will certain altered δAg forms that lack an RBD or a dimerization domain (Fig. 6A and B).
In summary this study used immunostaining of cells under defined conditions of δAg expression, without and with the accumulation of associated HDV RNA species, to examine the intracellular localizations of δAg and its colocalizations with host components. The findings reveal the complications of extrapolating from such localizations to interpretations of functional significance.
This work was supported by Public Health Service grants AI-26522 and CA-06927 and by an appropriation from the Commonwealth of Pennsylvania. C.A. was a recipient of a Fundação para a Ciência e Tecnologia Ph.D. grant.
Some confocal imaging was performed at the Biomedical Imaging Core, Department of Pathology and Laboratory Medicine, University of Pennsylvania, with the assistance of Xinyu Zhao. Other imaging was performed at the Fox Chase Imaging Facility under the direction of Michal Jarnik. Constructive comments on the manuscript were given by Ning Chai, William Mason, and Glenn Rall.
Published ahead of print on 15 April 2009.