![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Previous Article | Next Article 
Journal of Virology, May 2004, p. 4517-4524, Vol. 78, No. 9
0022-538X/04/$08.00+0 DOI: 10.1128/JVI.78.9.4517-4524.2004
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Fox Chase Cancer Center, Philadelphia, Pennsylvania 19111-2497
Received 12 November 2003/ Accepted 29 December 2003
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Ag-S, the small delta protein, was provided, the replication ability of the mutated RNA was restored. The second approach involved an examination of the processing in transfected cells of specific Pol II DNA-directed transcripts of HDV antigenomic sequences. The DNA constructs used were such that the RNA transcripts were antigenomic and began at the same 5' site as the mRNA produced during RNA-directed HDV genome replication. A series of such constructs was assembled in order to test the relative abilities of the transcripts to undergo processing by polyadenylation or ribozyme cleavage at sites further 3' on a multimer of HDV sequences. The findings from the two experimental approaches led to significant modifications in the rolling-circle model of HDV genome replication. |
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Ag-S (10). This 195-amino-acid protein is essential for HDV replication (6). During replication, the termination codon on the mRNA for Ag-S is changed as a result of posttranscriptional RNA editing by an adenosine deaminase (18), leading to the translation of a longer protein, large delta antigen, Ag-L, which may or may not act as an inhibitor of HDV replication (6, 19) but which is certainly needed for particle assembly (3). From this account, it should be clear that multiple posttranscriptional RNA-processing events must occur during the replication of HDV RNA. Previous studies have focused on these processing events (13, 21). Also, detailed double-rolling-circle models have been formulated to explain what might be the coordination of these processing events (9, 20, 21, 25). However, in parallel with consideration of such models, some attention must be given to the mechanism of RNA-directed transcription, especially since we know that some forms of RNA processing are associated with transcripts made by a particular RNA polymerase. Specifically, poly(A) processing is known to occur only for polymerase II (Pol II) transcripts.
In one rolling-circle model of HDV replication, it is presumed that all RNA transcription is carried out by the same RNA polymerase (9, 25). Two independent studies agree with the interpretation that both the mRNA and the genomic RNA arise via transcription using the host RNA Pol II (11, 20, 22). However, one study using the inhibitor 
-amanitin has been interpreted as evidence that the antigenomic RNA is transcribed by a polymerase other than Pol II (20). This interpretation, which we consider not to be justified by the data, has become the basis of a model in which the genomic RNA is somehow used as a template by two different polymerases: by Pol II, to make an RNA that is processed by 5' capping and 3' polyadenylation to produce the mRNA, and in addition by Pol I, to make an RNA that is ribozyme processed to produce the unit-length antigenomic RNA (20, 21).
Figure 1 presents the relevant part of a detailed rolling-circle model of HDV genome replication, in which all antigenomic RNA transcripts were assumed to be initiated by the same polymerase, presumably Pol II, and predominantly at a single site, namely, that corresponding to the 5' end of the what becomes a 5'-capped and 3'-polyadenylated mRNA (9-11). This model implies that such nascent transcripts could "continue" to be transcribed long distances beyond the poly(A) signals, so as to become greater-than-unit-length antigenomic species. These multimers could be processed by the antigenomic ribozyme to become unit-length linear RNAs which, in turn, would be ligated to produce unit-length circles. The studies described here were undertaken to test some of the predictions and speculations associated with this model. One of our findings is that RNA processing by polyadenylation and ribozyme cleavage should be considered events that in most cases act on two separate RNA transcripts. That is, a nascent RNA transcript may undergo either polyadenylation or ribozyme cleavage, but not both.
|
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-293, that expresses the small form of delta antigen, Ag-S. Transfections with either DNA or RNA were performed by using Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen).
Plasmids.
For the studies for which results are shown in Fig. 2, we used RNAs transcribed from plasmids pTW108 and pXN102. As previously described, construct pTW108, when opened with HindIII and transcribed in vitro with T7 polymerase, yields a genomic HDV RNA (29). This RNA is 1.2 times the unit length and has a 2-nt deletion in the open reading frame (ORF) for Ag-S; therefore, it is unable to make a functional protein. The RNA transcribed from pXN102 was identical to that from pTW108 except for the modification of the poly(A) signal from AAUAAA to UUUAAA.
|
Ag-S ORF were derived from plasmid pDL448, as previously described (15). Those with the frameshift in the ORF contained a 2-nt deletion at the unique EcoRI site, as previously described (6). The mutation of the poly(A) signal has also been described previously (13). These eight cDNA constructs were cotransfected into Huh7 cells along with pSVTVA, a plasmid which expresses both T-antigen and VA genes and enhances in vivo transcription from SV40-based vectors 16-fold (1).
|
|
Northern blot analysis.
Total RNA was extracted with Tri Reagent (Molecular Research Center), glyoxalated, and typically loaded onto 1.5% agarose gels. Electrophoresis and electrotransfer were performed as described previously (5). HDV RNA probes were prepared by T7 transcription with [-32P]UTP (Perkin-Elmer). Hybridization was performed at 65°C in Ekono Hybridization Solution (Research Products International). Radioactivity was detected and quantitated with a bio-imager (Fuji).
Selection of 5'-capped RNAs. For Fig. 2C, total RNA from transfected cells was fractionated to separate capped and noncapped species by using an affinity column to which was bound a monoclonal antibody that binds cap structures (2). First, this antibody was complexed to protein A-coated Sepharose (Sigma); then 50 µg of total RNA was applied. Bound RNAs were extracted with Tri Reagent. Bound and unbound RNAs were collected by ethanol precipitation prior to Northern blot analyses.
Selection of 3'-polyadenylated RNAs. For Fig. 4B, total RNA from transfected cells was fractionated to separate polyadenylated and nonpolyadenylated species. The strategy was essentially that described for the Dynabeads mRNA purification kit (Dynal). (dT)25 with a 5' biotin was first complexed to avidin-coated superparamagnetic beads. Then 8 µg of total RNA was subjected to binding and elution and was subsequently collected by ethanol precipitation prior to Northern blot analysis.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Therefore, we initiated HDV genome replication by using, instead of DNA constructs, greater-than-unit-length genomic RNAs, with and without the poly(A) signal mutation. These RNAs were transcribed in vitro. Both were first modified to contain a frameshift mutation in the ORF for the small delta protein. To provide a stable source of small delta protein, we carried out the transfection into a line of -293 cells that were able to express about 2.5 x 106 copies of this protein per cell in the presence of tetracycline. At day 2 after transfection, total RNA was extracted and examined by Northern blot analysis to detect antigenomic RNAs. Results are shown in Fig. 2A. We observed that for both wild-type and poly(A)-mutated genomic RNAs (Fig. 2A, lanes p and q, respectively), transfection led to significant accumulation of newly synthesized unit-length antigenomic RNA, which could be identified relative to a unit-length DNA standard in lane r. We also detected two polyadenylated mRNA species. As indicated in Fig. 2A, one was that stably produced by the -293 cells and the other was that produced by the replicating HDV genome. Note that for the poly(A)-mutated genome, much less of this mRNA was produced during HDV genome replication.
Figure 2B provides quantitation of the Northern blot data in Fig. 2A. The amount of each of the three RNA species in Fig. 2B, lanes q, is expressed relative to the amount in the corresponding lane p, taken as 100%. Note first that under these conditions the amount of unit-length RNA produced by the poly(A) mutant was 75% that of the wild type. In contrast, the amount of mRNA produced by this replication was only 14%, consistent with the mutation interfering with polyadenylation. As expected, the amount of mRNA from -293 cells was not changed. Similar results were obtained for RNAs harvested at days 3 and 4 after transfection (data not shown).
These results support the interpretation that as long as Ag-S was provided separately, the poly(A) mutation did not have an effect on unit-length viral RNA accumulation. Specifically, it did not have a major effect on the accumulation of these ribozyme-processed RNA transcripts. This finding is important, because Pol II is needed both for transcription and for several different aspects of RNA processing, and thus the poly(A) mutation could have had an effect on RNA transcription (27).
The studies for which results are shown in Fig. 2A provided a unique opportunity to compare and contrast the mRNA species that arose via RNA-directed versus DNA-directed transcription. Therefore, in the experiment for which results are shown in Fig. 2C, we tested the abilities of these two RNAs, present in total RNA isolated from a transfected culture as in Fig. 2A, lane p, to be selected for a 5'-cap structure. We used an affinity column with an antibody specific for the cap structure (2). As shown in Fig. 2C, we found that these two mRNAs were selected with significant and equal efficiency. In contrast, unit-length HDV RNA, used as a negative control, was less efficiently selected. This study thus provided the first direct evidence that the 5' end of RNA-directed HDV mRNA, like that of DNA-directed mRNA, is posttranscriptionally modified to contain a similar 5'-cap structure. The results confirm the previous evidence obtained by an indirect method (10).
Analysis of processing of DNA-directed antigenomic HDV RNA sequences. For the following studies, we devised a strategy to monitor the processing of nascent Pol II-directed transcripts of antigenomic RNA. As indicated in Fig. 3A, our aim was to monitor the processing events for RNAs that started at what we consider might be the 5' end for HDV RNA transcripts (10). Also, we wished to consider RNAs that could extend beyond unit length and might mimic the antigenomic transcripts achievable in rolling-circle replication. We realized that amounts of such transcripts that would allow characterization of the subsequent RNA processing could not be obtained in vivo from RNA-directed transcription. Furthermore, for these studies we actually wanted to study RNA processing in the absence of any RNA-directed replication. And because we wanted to consider the processing of HDV RNA transcripts of different lengths, with and without specific prior mutations, we could do this only in the context of DNA-directed transcription. With this in mind, we designed a multimeric HDV cDNA insert for an expression vector. This prototypic construct was such that when it was transiently transfected into cells, Pol II-directed RNA transcription would begin very close to the site corresponding to the 5' end of the HDV mRNA and then, as indicated in Fig. 3A, proceed through the ORF for the delta protein, the poly(A) signal, AAUAAA, the poly(A) acceptor site, CA, and the almost adjacent antigenomic ribozyme sequence. As indicated, this construct would next direct the transcription of a complete unit length of HDV antigenomic RNA, which again contained the ORF, the poly(A) signals, and a ribozyme.
Also shown in Fig. 3A are six HDV-specific RNAs that might possibly be derived by RNA processing from this multimeric antigenomic RNA transcript. These species are arranged in order of increasing size. Actually, they are organized as three pairs of transcripts; for each pair, the 3' end of one transcript is produced by polyadenylation (pA) while that of the other is produced by ribozyme cleavage (Rz). Note that species 1-pA and 2-Rz correspond to the natural HDV mRNA and unit-length antigenomic RNA, respectively.
Next, as shown in Fig. 3B, we considered the processing of RNA transcripts made not just from one but from a total of eight different constructs. The strategy was to compare and contrast the processing of different HDV RNA species. The first four constructs, which acted as templates for transcripts a through d, contained less-than-unit-length antigenomic RNA, while the last four, e through h, acted as templates for transcripts that were greater than unit length. In addition to changing the lengths of the transcripts, we employed mutagenesis of the poly(A) signals (transcripts c, g, and h) or the ribozyme domain (d). For transcripts b and f, we added an intron to the 5' sequences in order to determine whether this would have an effect on the processing of the HDV sequences.
In all of the eight processing studies that will now be described, we did not want the presence or absence of HDV RNA-directed replication to be a variable. We also wanted to exclude any effects on RNA processing that might be mediated by the presence or absence of the small delta protein. Therefore, we mutated the ORFs present in the RNA transcripts such that functional small delta protein was not expressed and there was no HDV replication.
In all of the following experiments, Huh7 cells were transfected with expression constructs, and at day 3, total RNA was extracted and subjected to Northern blot analyses to detect the processed HDV-specific sequences.
Consider first the analyses of the processing of transcripts a through d, which were less than unit length. Because of this short length, we could expect processed RNAs of class RNA-1 only, as indicated in Fig. 3A. The Northern blotting results are presented in Fig. 4A, and the position of RNA-1 is indicated. Note that in all lanes we detected a level of cross-hybridization to abundant host cell RNAs, especially 28S rRNA, as indicated for RNA from untransfected cells (Fig. 4A, lane u). Quantitation of the RNA-1 signal, after normalization to the 28S rRNA signal, is presented in Fig. 4B. The signals for transcripts b through d are expressed relative to that for transcript a, taken as 100%.
For transcript a, the band indicated as RNA-1 was heterogeneous and had an average size of about 800 nt, as expected for HDV mRNA (12). For these and other reasons, including poly(A) selection (data not shown), we identified this band as the polyadenylated species RNA-1-pA from Fig. 3A, and we believe that there was not a significant amount of RNA-1-Rz.
For RNA transcript b, a sequence containing an intron was added to the 5' noncoding sequence. Our intent with this construct was to determine whether the presence of an intron could change the nature or extent of RNA processing. Others have reported that poly(A) processing of host RNAs can be facilitated by the presence within the RNA precursor of intronic sequences (17). As shown in Fig. 4A, the band detected was less heterogeneous and also migrated somewhat more slowly. While the reduced heterogeneity may be due in part to a more uniform length of polyadenylation, we consider the increased size to be due largely to the 276 nt of added non-HDV sequences remaining after the splicing out of the 973-nt intron. Interestingly, from quantitation of the Northern blot data, we found that the amount of HDV-specific signal was not detectably different from that for transcript a. Thus, we conclude that processing of the normal HDV mRNA is not at any significant disadvantage relative to that for an RNA with an intronic sequence.
RNA transcript c had the mutant poly(A) signal, UUUAAA. As expected, the amount of RNA-1-pA was reduced significantly, to 14% that of transcript a.
In transcript d, the ribozyme located 3' of the poly(A) signals was deleted. Nevertheless, the amount of RNA-1 was only changed 30% from that detected for transcript a. This result indicates that the presence of the ribozyme domain is not needed for polyadenylation. It also shows that the bands indicated here as RNA-1 are fully RNA-1-pA and do not contain any species that could be RNA-1-Rz.
Next consider the processing of RNA transcripts e through h, which differ from transcripts a through d in that a complete unit-length HDV sequence has been added 3' of the poly(A) signals and the first ribozyme. Therefore, we might now be able to detect some or all of the six processed RNA species diagrammed in Fig. 3A. In the Northern blot results (Fig. 5A), we detect the mass distribution of species RNA-1 to RNA-3. In order to determine the relative molar distribution, we first normalized the hybridization signals to the amount of the cross-hybridizing band of 28S rRNA and then corrected for the length of the HDV-specific sequence present in each species. We thus obtained the relative molar distribution of the three pairs of processed RNA species for each of RNA transcripts e through h (Fig. 5D).
|
For transcript f, there was only one change relative to transcript e, namely, the presence of the 5' intron. In terms of the resultant RNA processing, the only obvious consequence was the reduced mobility of RNA-1. We conclude that just as for transcript b relative to transcript a, the presence of the intron in the nascent transcript led to a processed polyadenylated species, RNA-1-pA, that was less heterogeneous and also somewhat larger. A second but less obvious change relative to transcript e was that the slowest-migrating band, indicated as RNA-3, was less heterogeneous in size. In order to better characterize this species, the total RNA was fractionated by poly(A) selection. As shown in Fig. 5C, we fractionated the total RNA into nonpolyadenylated and polyadenylated RNAs. It can now be seen that RNA-3 was predominantly polyadenylated and can thus be identified as RNA-3-pA rather than RNA-3-Rz. Likewise, the RNA-1 band can be identified as predominantly RNA-1-pA. The RNA-2 band is identified as RNA-2-Rz.
For transcript g, the first poly(A) signal has been removed. As expected, this led to a reduced amount of RNA-1-pA. In addition, it increased the amount of RNA-3. However, after quantitation and normalization of the data, as shown in Fig. 5D, there was no significant change in the amount of RNA-2-Rz.
In transcript h the second poly(A) signal was mutated. The only obvious consequence for RNA processing was a major reduction in the amount of RNA-3. This is expressed clearly in Fig. 5D. The residual RNA-3 was not polyadenylated (data not shown) and is identified as RNA-3-Rz. It is important that this mutation of the second poly(A) signal had no detectable effect on the levels of RNA-1 and RNA-2.
As considered further in the Discussion, the results of this examination of RNA processing for the eight different RNA transcripts led us to propose the interpretation that poly(A) addition and ribozyme cleavage to a large extent are alternative processing outcomes for separate antigenomic HDV RNA transcripts.
While Ag expression is not directly related to the question of RNA processing, we performed, in parallel with the Northern blot analyses for which results are shown in Fig. 4A and 5A, immunoblot assays to detect Ag. Two points were revealed from this analysis (data not shown). First, for both transcripts c and g, mutation of the first poly(A) signal led to a 50 to 60% reduction in the accumulation of Ag. Second, for both transcripts b and f, insertion of the intron in the 5' untranslated region produced a 70 to 100% increase in Ag levels. Since the amounts of RNA-1 produced by transcripts b and f were unchanged, it may be that the involvement of RNA splicing led to more efficient mRNA transport from the nucleus and subsequent translation in the cytoplasm. Such results are consistent with the reports of others that exon junction complexes formed on an RNA can enhance the expression of protein from the mRNA (16, 17, 28).
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The double-rolling-circle model of replication shown in Fig. 1 suggests that antigenomic transcripts, once initiated, could first be processed to yield capped and polyadenylated mRNA, and then the "continuing transcript," following further elongation, could be processed by two ribozyme cleavages to release unit-length RNAs, which in turn could be processed by ligation to become unit-length circular antigenomic RNAs (13, 25). In a contrasting model, Modahl and Lai (21) have asserted that poly(A) processing and ribozyme cleavage are completely separate processing events, because the substrate for each is transcribed by a different RNA polymerase. While they agreed that Pol II transcribes the species that becomes the mRNA, they asserted that a different polymerase, suggested to be Pol I, is used for those RNAs which become the unit-length antigenomic RNA (20). The present study clearly shows that of RNAs synthesized in vivo using the same template sequence and directed by the same Pol II promoter, some are processed by polyadenylation and others are processed to become unit-length RNA circles.
In agreement with the concept of mutually exclusive processing, we found that mutating the poly(A) signal had no effect on the level of ribozyme processing (Fig. 5A and D, lanes g and h). Conversely, removal of a ribozyme had no effect on the level of polyadenylation (Fig. 4, lanes d).
It is relevant that since the earlier rolling-circle model was first formulated, a greater understanding of the process of polyadenylation has been achieved. Dye and Proudfoot have shown that nascent RNAs often proceed 500 to 3,000 nt beyond the poly(A) signal before transcriptional termination can occur, followed by disengagement of the polymerase from the template, formation of a complex of polyadenylation factors at and around the poly(A) signal, cleavage at the poly(A) acceptor site, and finally addition of poly(A) to that site (8). Therefore, we now suggest that for HDV it is more likely that the "choice" between poly(A) processing and ribozyme cleavage not only is stochastic but also is made after nascent transcripts have achieved sizes much greater than unit length.
This interpretation of mutually exclusive processing then leads to the question of what factors determine this choice. It seems that there should be no difference in the initiation of transcription. It could well be that at some point(s) after transcription is initiated, differences in the recruitment of host factors needed for the two processing events lead to the alternatives in processing.
In the earlier rolling-circle model of HDV replication, it was proposed that a viral protein, Ag-S, could bind to nascent RNAs that can fold into the rod-like structure and preferentially suppress processing by polyadenylation. This was asserted largely because Hsieh and Taylor had shown that Ag-S could inhibit polyadenylation (13). However, in contrast to the present studies, Hsieh and Taylor did not have a simultaneous measure of polyadenylated RNAs and RNAs that were processed to become unit-length antigenomic RNAs. Furthermore, the present studies show that the added presence during transcription of either Ag-S or Ag-L was not able to decrease the level of poly(A) processing (data not shown).
During HDV replication either in the liver of an infected animal or in cultured cells, the molar amount of polyadenylated mRNA is often 50 times less than that of the unit-length antigenomic RNA species produced by ribozyme cleavage (7, 10, 12). In contrast, when we determined the molar amounts of the main forms of processed DNA-directed RNA transcripts, we observed that the short polyadenylated mRNA species was three times more abundant than the unit-length RNA (Fig. 5D). This ratio was not significantly changed when either Ag-S or Ag-L was transiently expressed, and it was not changed in -293 cells when Ag-S was provided by stable transfection (data not shown). One possible interpretation is that somehow the nature of our transient expression studies offered less opportunity for the decay of mRNA than for that of RNA circles. Consistent with this explanation was the fact that even species polyadenylated at a second site were detected in these experiments, and yet such RNAs have never been reported during HDV genome replication in the liver or in cultured cells.
In summary, the present studies demand that significant changes be made to the rolling-circle models of HDV genome replication (Fig. 1) (9, 20), especially as they might apply to the processing of antigenomic RNA transcripts. First, according to these models, polyadenylation is needed to make the mRNA for the small delta protein, but the new data make clear that if the protein itself is provided separately, neither the presence nor the functioning of the polyadenylation signals is needed for HDV replication. Second, contrary to the model, ribozyme cleavage of antigenomic RNAs neither precedes nor is necessary for the ability of such RNAs to undergo poly(A) processing. Third, the present studies show that transcripts starting at the same location and using the same polymerase (Pol II) can be processed to yield either mRNA or unit-length RNA circles. (Incidentally, contrary to one published model [20], two different polymerases are not needed.) A concept of the old model is that a nascent antigenomic RNA undergoes first poly(A) processing and then ribozyme processing. However, this concept is not necessary and probably is incorrect, because in the present study, the two forms of processing appear to be largely mutually exclusive. That is, it is more likely that a given nascent HDV transcript will undergo either one form of processing or the other, but not both. To some extent such alternative possibilities are analogous to what we realize happens for most mRNA precursors that contain multiple introns; such RNAs can undergo multiple alternatives in terms of which introns are spliced (24).
In conclusion, the results presented here, along with other data recently reviewed (26), make clear that HDV genome replication is much more complex than the rolling-circle models have led us to believe.
|
ACKNOWLEDGMENTS |
|---|
This work was supported by grants AI-26522 and CA-06927 from the NIH and by an appropriation from the Commonwealth of Pennsylvania.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
antigen is crucial for the assembly of hepatitis virus. Proc. Natl. Acad. Sci. USA 88:8490-8494. virus. Proc. Natl. Acad. Sci. USA 83:8774-8778.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| J. Bacteriol. | Mol. Cell. Biol. | Microbiol. Mol. Biol. Rev. |
|---|
| Clin. Vaccine Immunol. | ALL ASM JOURNALS |
|---|
