Evolution of Hepatitis Delta Virus RNA Genome foll
http://www.100md.com
病菌学杂志 2005年第21期
Fox Chase Cancer Center, 333 Cottman Avenue, Philadelphia, Pennsylvania 19111-2497
ABSTRACT
Previous studies have defined a novel cell culture system in which a modified RNA genome of hepatitis delta virus (HDV) is able to maintain a low level of continuous replication for at least 1 year, using a separate and limited DNA-directed source of mRNA for the essential small delta protein. This mode of replication is analogous to that used by plant viroids. An examination was made of the nucleotide changes that accumulated on the HDV RNA during 1 year of replication. The length of the RNA genome was maintained, except for some single-nucleotide deletions and insertions. There was an abundance of single-nucleotide substitutions, with a 22-fold excess of these being base transitions rather than transversions. Of the detected transitions, at least 70% were consistent with being the consequences of posttranscriptional RNA editing by an adenosine deaminase acting on RNA. The remainder of the changes, including the single-nucleotide insertions and deletions, are likely to be the consequence of misincorporation during transcription. In addition, an intermolecular competition assay was used to show that the majority of the genomes present after 1 year of replication were essentially as competent in replication as the original single HDV RNA sequence that was used to initiate the genome replication. A model is provided to explain how, in this experimental system, the observed single-nucleotide changes were essentially neutral in terms of their effect on the ability of the HDV genome to carry out continued rounds of replication.
INTRODUCTION
Previous studies have considered the nucleotide changes that accumulate during the replication of the approximately 1,700-nucleotide single-stranded circular RNA genome of human hepatitis delta virus (HDV) (1, 12, 15, 22, 25, 28, 37). Some of these changes are considered to arise via misincorporations during RNA-directed RNA transcription, a process that is considered to involve redirection of host DNA-directed RNA polymerase II (34). In contrast, one much-studied nucleotide change that occurs at position 1012, in the middle of the amber termination codon of the open reading frame for the essential small delta protein, Ag-S (23), is not caused by misincorporation. This change occurs during genome replication but is a consequence of posttranscriptional RNA editing by ADAR, an adenosine deaminase acting on RNA (2). This change leads to inactivation of the termination codon and allows the synthesis of a longer protein, the 214-amino-acid large delta protein, Ag-L, that is a dominant negative inhibitor of genome replication (6) and yet is essential for the process of virus assembly which is mediated by the envelope proteins of a helper virus, hepatitis B virus (3). There are other less-site-specific nucleotide changes that have been noted in previous studies, whether during passage of the virus in infected animals or even during replication in cultured cell lines (12, 25, 28). It is no doubt because of the accumulation of changes such as these that the different HDV isolates from around the world can be divided into at least three, and possibly seven, different genotypes (29).
In the present studies, we have made use of a recently described novel system for HDV genome replication in order to follow the accumulation of changes that occur during extended replication of the HDV genome (4). Briefly, the system uses a line of human embryonic kidney cells, 293-HDV, that contain a single copy of a cDNA to express Ag-S, under tetracycline-on control. In addition, the cells have been transfected with an HDV RNA that by the creation of a 2-nucleotide deletion has disrupted the open reading frame for Ag-S (6). This genome, however, can replicate using the unchangeable source of Ag-S from the cDNA. If tetracycline (TET) is added, these cells produce large amounts of Ag-S, leading to extensive HDV replication, cell cycle arrest, and cell death. However, in the absence of TET, there is sufficient Ag-S translated to allow the mutated genome to replicate at the low level of about 1,000 copies per cell, without any indication of cytotoxic effects, for at least 1 year (4).
The strategy of the present study includes the use of reverse transcription-PCR (RT-PCR), cloning, and sequencing to detect and characterize the changes that occurred on the HDV RNA during 1 year of replication. In addition, we used a previously described competition assay to examine the replication competence of the HDV genomes at this time, relative to that of the single HDV RNA sequence that was first used to initiate the genome replication (13).
MATERIALS AND METHODS
Cell culture. The creation and maintenance of low-level HDV genome replication was carried out as previously described (4). Briefly, a line of 293-TREx cells (Invitrogen) was stably transformed with cDNA encoding the HDV small delta antigen under TET control. These cells, designated 293-Ag, were then transfected with an HDV antigenomic RNA containing a 2-nucleotide deletion in the open reading frame for the small delta antigen (6). When these cells were maintained under TET– conditions, a low level of HDV replication (about 1,000 copies per cell) was detected for periods of at least 1 year. These cells were designated 293-HDV.
RNA analysis. Unless specified otherwise, 293-HDV cells were induced with TET (1 μg/ml) for 2 days prior to harvest. Total RNA was extracted using Tri Reagent (Molecular Research Center, Inc.). RT-PCR, cloning, and nucleotide sequencing were carried out as previously described (5). The DNA primers used are described in Fig. 1. HDV RNAs were also examined by Northern analyses using 32P-labeled riboprobes, as previously described (38).
Intermolecular competition assay. This was a modification of a previously described assay that measures the relative replication competence of HDV genomes (13). It was carried out by cotransfection of new cells with a combination of two sources of total RNA. The first was from 293-HDV cells at 1 year. The other was from Huh7 cells at 3 days after transfection with wild-type HDV RNAs that contained a unique marker sequence.
As a preliminary to our competition experiments, we used Northern analyses to normalize the amounts of HDV RNA. Then, approximately equal amounts of the two RNAs were cotransfected into recipient cells. At various times thereafter, total RNA was extracted and assayed to measure the relative replication of these two forms of HDV RNA, using Northern analysis with oligonucleotide probes specific for the marked and unmarked sequences (13).
More specifically, we detected the genomes at 1 year with the following antigenomic oligonucleotide (nucleotides 804 to 784): 5'-CTCGATTCTCTA TCGGAATCT-3'. This oligonucleotide was chosen in a region that was absolutely conserved during 1 year of replication (see Table S1 in the supplemental material). In contrast, the marked wild-type genomes were detected with the oligonucleotide 5'-CTCGATTCTGCTTCAGAATCT-3'. This sequence differs from the wild-type sequences at the nucleotides that are underlined.
RESULTS
Strategy for RT-PCR, cloning, and sequencing. The following strategy was used for the cloning and sequencing of the HDV RNA genomes after 1 year of continuous low-level replication in the 293-HDV cell line. Two days prior to the harvest of total RNA, TET was added to the cells in order to transiently increase the copy number of HDV RNAs and thereby facilitate the subsequent HDV RNA analysis. (As mentioned below, control studies were performed, which showed that such induction for 2 days was not a significant source of nucleotide changes relative to the 1 year of replication in the absence of TET.) Four primer pairs, as indicated in Fig. 1, were used to perform reverse transcription and PCR amplification. These four regions have significant overlap and together represent the entire HDV genome sequence. These PCR products were cloned, and in each case, at least nine clones were subjected to nucleotide sequencing.
Detected sequence changes. Table 1 summarizes the nucleotide sequence changes detected. Figure 2 shows the location of each of these changes relative to the full HDV RNA sequence and the locations of known sequence features of both the genome and the antigenome. Complete information for all changes is presented in Table S1 of the supplemental material.
For 26,761 total nucleotides sequenced, we detected 313 changes at 121 different sites relative to the HDV sequence that was first used to initiate replication in these cells. This represents about 2.1% changes per nucleotide per year. This value seems high relative to the range of 0.001 to 0.34% for 50 different RNA viruses (17); however, as discussed below, we estimate that around 70% of our detected changes may have arisen not as cotranscriptional misincorporations but as posttranscriptional RNA editing events.
There were no major changes in the length of the HDV genome. The only length changes detected were of relatively rare single-nucleotide insertions and deletions. These, representing only 7% of all changes, occurred at several different locations. (In a previous study of passaged HDV, such changes were not detected, that is, represented <2.5% of all changes [25].)
The majority of the detected changes (93%) were single-nucleotide substitutions. Of these, the majority were base transitions rather than transversions. Furthermore, not all possible base transitions were observed with comparable frequencies. The CU and GA changes combined were less than 10%. In contrast, 90% of the changes were consistent with either a UC or an AG.
Two sites underwent nucleotide changes in 100% of clones sequenced. The first, located at position 1375, has been reported in one previous study and is probably an error due to a misincorporation during RNA-directed RNA synthesis (12). Now, from studies by us and by others it is established that the second change, at position 1012, arises via posttranscriptional RNA editing by an adenosine deaminase acting on RNA, an ADAR (2, 23). This editing converts AI on the antigenome and, with further replication, changes the genome from UC and the antigenome from AG. Knowing this, and knowing that other sites of putative ADAR editing have been shown on HDV RNAs (25), we would interpret that most of the UC or AG changes detected on the genomic sequence are the consequences of ADAR editing on either genomic or antigenomic RNAs. From this assumption, we deduce that about 70% (79.6 – 9.2%) of all of the detected changes could have arisen via ADAR editing, with 18% (9.2 + 9.2%) being base transitions due to misincorporation. Further support for this interpretation was gained by an examination of the 5' neighbors for these putative ADAR editing sites. That is, the frequencies for the observed 5'-neighboring nucleotides were very similar to those of the naturally occurring editing sites on RNAs with secondary structure (18) (see Table S2 in the supplemental material for more details.)
If as many as 70% of the changes could be due to ADAR editing and 7% are single-nucleotide insertions and deletions, the remaining 23% are probably the consequence of misincorporations during transcription. Three different changes were detected at position 553; this might suggest some template structure enhanced transcriptional errors. Similarly, at position 611 there was a frequent deletion of one nucleotide.
However, we do have to allow that some fraction of these detected changes may have arisen as a consequence of our in vitro analysis. We estimate the combined error rate for RT-PCR, cloning, and associated nucleotide sequencing to be <0.2% (12). With this in mind, in Fig. 2 we indicated changes that were detected only once. It can be noted that of the changes detected only once, 78% were UC or AG, which could be explained as ADAR editing. (See Table S1 in the supplemental material for more details.)
We carried out certain controls to determine whether the changes were accumulating with the extent of HDV replication. One control was to examine the replicating HDV RNA sequences at the earlier time of 16 weeks. This was done for the region flanked by the primer pair A, shown in Fig. 1. We definitely found changes at this time, especially at position 1012, and yet the extent was noticeably less than that discovered at 1 year. Thus, at 1 year we were detecting changes that accumulated with time of replication (data not shown). Another control was to determine whether significant changes arose during the brief 2-day period of TET induction that was used to amplify the intracellular HDV RNA levels, thus facilitating the RT-PCR strategy. We therefore compared the sequences of clones made from the region flanked by primer pair A (Fig. 1) for RNA from uninduced cells at 1 year relative to those after the 2 days of TET induction. Both contained the same spectrum of nucleotide changes (data not shown).
We also considered the detected sequence changes in terms of frequency and location relative to the sequence motifs that can be defined for the HDV genome and antigenome. Figure 2 is such a representation. It can be seen, as mentioned above, that the most frequently detected changes were at position 1012, within the termination codon for Ag-S, and at position 1375.
Several other sites were also abundant. Most striking was the cluster of changes in the region at and surrounding the poly(A) signal, AAUAAA. The changes in this region are shown in greater detail in Fig. 3, representing a total of 17 independent sequences. Strikingly, this cluster of changes (like that at position 1012) is consistent with ADAR editing, either on the antigenome or genome. To test whether these changes had an impact on HDV poly(A) processing, we used Northern analyses. As shown in Fig. 4, we observed not one but two species of HDV mRNA derived from the replication of these genomes. Relative to controls, we deduce that one species was as normally found during HDV genome replication while the second migrated somewhat more quickly. It is possible that the faster species contained a shorter poly(A) tail. Maybe the poly(A) signal changes on the genomes at 1 year interfered with addition of the poly(A) tail, for example, as an intrinsic fault during HDV poly(A) processing and/or an inability to maintain a typical full-length poly(A) tail.
In considering these data, it is important to keep in mind that in our experimental system, translation of Ag-S from the mRNA produced during HDV genome replication is no longer needed. Nevertheless, our data show that even after 1 year of replication and all the accumulated changes, one or more mRNA species continue to be processed. Since some form of poly(A) processing was still maintained, we are tempted to conclude that this processing, independent of the need for Ag, is somehow an essential part of the replication scheme.
Finally, we noted that some regions of the HDV sequence had relatively few, if any, changes. These regions include the two ribozyme domains which are needed for HDV RNA processing and known to be essential for HDV replication (16). Totally conserved was the end of the rod-like structure that we refer to as the top. This region is considered to include the site for the initiation of the HDV mRNA (10, 11). Another relatively conserved region is between positions 219 to 478; we currently have no explanation for this conservation.
Replication competence. Figure 5A summarizes the result of an intermolecular competition assay to determine the replication competence of the genomes at 1 year relative to that of the RNA used to initiate replication in the first place. This strategy was carried out largely as previously described (13). The assay involves cotransfection of recipient TET-induced 293-Ag cells with the two sources of HDV RNA, and at various times thereafter, total RNA is extracted and assayed by Northern analysis with oligonucleotide probes that can specifically detect the genomes at 1 year and the wild-type genomes. Any change in the ratio of these genomes is considered a measure of the relative ability to initiate and continue HDV replication. In the cotransfection, each cell receives multiple copies of each type of genome and they share the available Ag.
Three experiments were performed with the data for the average values and the range of these values as represented in Fig. 5A. As shown, the ratio of genomes at 1 year to wild-type genomes did not change significantly between day 0 and day 1. This means that the HDV RNAs at 1 year were just as competent as the wild type in initiating HDV genome replication. They were neither more nor less competent. Furthermore, this ratio did not change for times after 1 day. Therefore, the genomes at 1 year were not at any significant genetic advantage or disadvantage, since either of these could lead to a time-dependent change in the ratio.
We next asked whether the genomes at 1 year, even though they were fully replication competent, had actually been selected for a replication competence that was unique to our competition assay system that uses the 293-Ag cells during TET induction. We tested the ability of HDV at 1 year to compete with wild-type HDV RNA in 293 cells when the sources and/or amounts of available Ag were varied. First, we tested the competition in 293-Ag cells when there was no TET induction (Fig. 5B). Also, we tested competition in 293 cells when the only source of Ag was cotransfection of the relevant mRNA as first transcribed in vitro. In both cases, we found the genomes at 1 year to have replication competence comparable to that of the wild type (Fig. 5B).
However, a different answer was obtained when we tested replication competence in cells other than those related to 293 cells, such as Huh7 and HeLa cells. For these competition assays, the HDV RNAs were again cotransfected along with in vitro-transcribed mRNA to express the essential Ag-S. In both cases, there was a sixfold drop in the ratio (Fig. 5B) and these reduced ratios did not decrease further with time after transfection (data not shown). One interpretation is that, when assayed under these conditions, only 15% of the genomes at 1 year were as competitive as the wild type for initiation of HDV replication in these cells and that these genomes that did initiate replication were not at any genetic disadvantage relative to the wild type.
It was important to determine whether those genomes at 1 year that could replicate when passaged into fresh 293 or Huh7 cells in the absence of competitor wild-type virus maintained the many nucleotide sequence changes. Therefore, at 4 days after transfection, the total RNA was extracted and the HDV sequence flanked by the primer pair A in Fig. 1 was subjected to RT-PCR, cloning, and sequencing. We observed that the sequences obtained still contained many single-nucleotide changes, comparable to the genomes at 1 year that were used to initiate this additional round of replication (data not shown). Thus, while we do favor that there was a selection associated with replication in different host cells, the data argue against an obvious selection for a minor population of HDV genomes, for example, with relatively fewer sequence changes.
DISCUSSION
The present experiments have provided new and important information on the evolution of replicable RNA molecules over time. Previously, others have tried to follow the evolution of RNA molecules that can be replicated in the test tube. Two important examples of this are studies of the phage Q? mini-replicon (27, 33) and the short RNAs replicated by a phage RNA polymerase (19, 20). In our studies, the HDV RNA genome was replicated in a cell line that had been transfected with RNA 1 year earlier. There was no cell-to-cell spread of these HDV genomes other than at what must be about 365 consecutive cell divisions. Throughout all of this, the HDV replication was totally dependent upon an integrated cDNA source of fully functional, unchanging Ag-S. In this way, the cells maintained about 1,000 copies of HDV genomic RNA per cell.
We consider that at least two negative selective pressures acted against the survival of replicable HDV RNA in our experimental system. The first was the accumulation of errors during RNA-directed RNA transcription. The second was consistent with posttranscriptional RNA editing by ADAR activity. Such editing altered virtually all of the RNAs at position 1012, but there were in addition many single-nucleotide changes that were of the specific form consistent with ADAR editing of either genomic or antigenomic RNAs. From Table 1, we might thus deduce that among the RNA genomes that accumulated, ADAR editing was responsible for more changes that were tolerated than was nucleotide misincorporation. Other than the present study, there are two published reports where HDV nucleotide sequences changes were followed after beginning replication with a single HDV sequence (25, 28). Both studies examined only about 350 nucleotides of the HDV genome. In one study, HDV was serially passaged by infection of woodchucks, using woodchuck hepatitis virus as the helper virus. In the other study, the sequence changes were detected following a transfection with cDNA. In both studies, like in the present study, at least 70% of the single-nucleotide changes observed could have been caused by ADAR editing.
While nucleotide misincorporation during transcription and ADAR editing acting posttranscription might have been negative pressures acting on HDV genome replication, as now explained, we consider that there was also a positive selective pressure to maintain genome replication. In our experimental system, each cell contains about 1,000 copies of HDV genomic RNA. As each cell divides, another 1,000 copies need to be transcribed, processed, and accumulated. If a fraction of the original 1,000 copies had been compromised by the above-mentioned negative pressures and become less replication competent, then most of the new 1,000 copies would be made from the noncompromised genomes. This purging effect would be achieved with every cell division over the 1-year period of study. Thus, without the wisdom of hindsight, we should have expected that after 1 year, the surviving genomes would all be just as replication competent as the original wild-type sequence. In addition, those that had undergone any nucleotide changes would be such that the changes were totally neutral and not even detectably compromised in replication competence relative to the wild type. In retrospect, we also realize that if any of the changed genomes were able to replicate not less but more efficiently, this in turn might also be selected against. After all, as previously reported, enhanced replication of HDV RNA in this cell system can lead to cell cycle arrest followed by cell detachment and death (4).
Since all of the genomes sequenced at 1 year contained about 20 nucleotide changes, many of which were at diverse locations, we conclude that these were changes that did not detectably interfere with replication competence. That is, these genomes contained all of the necessary cis-acting sequences and structural features essential for uncompromised replication (Fig. 2).
The nature of our experimental system is that these conserved features do not include maintaining the ability to make mRNA that can be translated to make the essential Ag-S. Thus, in our experimental system HDV genome replication is mimicking that replication which is achieved by viroids. These are infectious subviral agents of plants with small, single-stranded, circular RNA genomes whose replication is totally dependent on host-encoded proteins (8).
One prediction of the above interpretation of a positive selective pressure is that if a change arises which compromises genome replication, there may not be an opportunity for the appearance of a second-site change that might rescue the genome. If the only changes that can accumulate with time are those that, per se, are totally neutral, then to explain the observed total of about 20 changes per genome, we predict that these must have accumulated over time from a series of single-nucleotide changes, each of which was totally neutral.
For many RNA viruses, there has been extensive thinking and some controversy, in terms of the concept of quasi-species, about which heterogeneous populations of viral genomes can provide mutual support to achieve a level of replication (7). Maybe this concept is not applicable to HDV replication as it is examined in our studies. This is because none of the genomes that are replicating can make a functional Ag-S. All of the genomes are dependent upon a separate source of this Ag-S. However, if the HDV replication were not supported by an external source then the quasi-species model could apply to some extent. That is, if some of the viral genomes were providing a source of functional Ag-S, they would be able to support, in trans, the replication of other genomes that were unable to make any Ag or that made a Ag-S that was not fully functional. Of course, if the supported genomes were making a significant amount of a form of Ag-S that interfered with HDV replication (like Ag-L), then they may compromise the coreplication. It should also be noted that some aspects of the quasi-species models demand that the replicating genomes can undergo significant levels of intermolecular recombination. For HDV, this has been claimed (35, 36), but the issue is still controversial (14, 34).
What about epistasis? Recent studies have addressed the issue of whether this can contribute to the fitness of an RNA virus (9, 24, 30-32). The discussion has included concepts of both positive and negative epistasis. It might be considered that our studies do not impact this question because we do not have clones for individual full-length HDV RNAs. One would have to obtain such clones and test them individually for their ability to initiate HDV replication. However, our evidence that the major fraction of the genomes at 1 year are as replication competent as the wild type, together with the observation that the majority have also undergone sequence changes, supports the interpretation that such changes have not conspired to produce combinations of positive and negative epistatic effects. At the same time, we cannot exclude that deliberately introduced changes might not achieve such effects.
Finally, we come to the question of what happens in terms of replication fitness when the HDV RNAs at 1 year are transfected to initiate replication in cells other than the 293 cells in which they were maintained for 1 year. We found from our competition assays that in other human cell lines, whether of liver or nonliver origin, only about 15% of the genomes had a replication competence that was equivalent to that of wild-type genomes (Fig. 5B). What then was the disadvantage for the 85% that were unable to initiate replication? Maybe these altered genomes were of a secondary structure that could not be adequately protected by Ag-S when expressed in Huh7 and HeLa cells.
In summary, the present study has been able to exploit a new model of HDV replication to examine the long-term consequences of such replication on the integrity of the HDV genomes. The findings have significant implications in terms of (i) the conservation of essential cis-acting sequence features, (ii) the fidelity of RNA-directed RNA synthesis by a host polymerase, and (iii) the impact of ADAR editing.
ACKNOWLEDGMENTS
This work was supported by grants AI-26522 and CA-06927 from the NIH and by an appropriation from the Commonwealth of Pennsylvania.
We thank Anita Cywinski and the Fox Chase DNA Sequencing Facility (Randy Hardy, manager). Constructive comments on the manuscript were given by Sven Behrens and Richard Katz. We thank Raul Andino for pointing out the need to test replication fitness in multiple cell types.
Supplemental material for this article may be found at http://jvi.asm.org/.
REFERENCES
Casey, J. L., T. L. Brown, E. J. Colan, F. S. Wignall, and J. L. Gerin. 1993. A genotype of hepatitis D virus that occurs in northern South America. Proc. Natl. Acad. Sci. USA 90:9016-9020.
Casey, J. L., and J. L. Gerin. 1995. Hepatitis D virus RNA editing: specific modification of adenosine in the antigenomic RNA. J. Virol. 69:7593-7700.
Chang, F. L., P. J. Chen, S. J. Tu, M. N. Chiu, C. J. Wang, and D. S. Chen. 1991. The large form of hepatitis antigen is crucial for the assembly of hepatitis virus. Proc. Natl. Acad. Sci. USA 88:8490-8494.
Chang, J., S. O. Gudima, C. Tarn, X. Nie, and J. M. Taylor. 2005. Development of a novel system to study hepatitis delta virus genome replication. J. Virol. 79:8182-8188.
Chang, J., and J. Taylor. 2002. In vivo RNA-directed transcription, with template switching, by a mammalian RNA polymerase. EMBO J. 21: 157-164.
Chao, M., S.-Y. Hsieh, and J. Taylor. 1990. Role of two forms of the hepatitis delta virus antigen: evidence for a mechanism of self-limiting genome replication. J. Virol. 64:5066-5069.
Domingo, E. 2000. Viruses at the edge of adaptation. Virology 270:251-253.
Flores, R., S. Delgado, M. E. Gas, A. Carbonell, D. Molina, S. Gago, and M. De la Pena. 2004. Viroids: the minimal non-coding RNAs with autonomous replication. FEBS Lett. 567:42-48.
Froissart, R., D. Roze, M. Uzest, L. Galibert, S. Blanc, and Y. Michalakis. 2005. Recombination every day: abundant recombination in a virus during a single multi-cellular host infection. PLoS Biol. 3:e89.
Gudima, S., K. Dingle, T.-T. Wu, G. Moraleda, and J. Taylor. 1999. Characterization of the 5' ends for polyadenylated RNAs synthesized during the replication of hepatitis delta virus. J. Virol. 73:6533-6539.
Gudima, S., S.-Y. Wu, C.-M. Chiang, G. Moraleda, and J. Taylor. 2000. Origin of the hepatitis delta virus mRNA. J. Virol. 74:7204-7210.
Gudima, S. O., J. Chang, G. Moraleda, A. Azvolinsky, and J. Taylor. 2002. Parameters of human hepatitis delta virus replication: the quantity, quality, and intracellular distribution of viral proteins and RNA. J. Virol. 76: 3709-3719.
Gudima, S. O., J. Chang, and J. M. Taylor. 2004. Features affecting the ability of hepatitis delta virus RNAs to initiate RNA-directed RNA synthesis. J. Virol. 78:5737-5744.
Gudima, S. O., J. Chang, and J. M. Taylor. 2005. Reconstitution in cultured cells of replicating HDV RNA from pairs of less than full-length RNAs. RNA 11:90-98.
Imazeki, F., M. Omata, and M. Ohto. 1990. Heterogeneity and evolution rates of delta virus RNA sequences. J. Virol. 64:5594-5599.
Jeng, K.-S., A. Daniel, and M. M. C. Lai. 1996. A pseudoknot structure is active in vivo and required for hepatitis delta virus RNA replication. J. Virol. 70:2403-2410.
Jenkins, G. M., A. Rambaut, O. G. Pybus, and E. C. Holmes. 2002. Rates of molecular evolution in RNA viruses: a quantitative phylogenetic analysis. J. Mol. Evol. 54:156-165.
Knight, S. W., and B. L. Bass. 2002. The role of RNA editing by ADARs in RNAi. Mol. Cell 10:809-817.
Konarska, M. M., and P. A. Sharp. 1989. Replication of RNA by DNA-dependent RNA polymerase of phage T7. Cell 57:423-431.
Konarska, M. M., and P. A. Sharp. 1990. Structure of RNAs replicated by the DNA-dependent T7 RNA polymerase. Cell 63:609-618.
Kuo, M. Y. P., J. Goldberg, L. Coates, W. Mason, J. Gerin, and J. Taylor. 1988. Molecular cloning of hepatitis delta virus RNA from an infected woodchuck liver: sequence, structure, and applications. J. Virol. 62: 1855-1861.
Lee, C. M., F. Y. Bih, Y. C. Chao, S. Govindarajan, and M. M. C. Lai. 1992. Evolution of hepatitis delta virus RNA during chronic infection. J. Virol. 188:265-273.
Luo, G., M. Chao, S.-Y. Hsieh, C. Sureau, K. Nishikura, and J. Taylor. 1990. A specific base transition occurs on replicating hepatitis delta virus RNA. J. Virol. 64:1021-1027.
Michalakis, Y., and D. Roze. 2004. Evolution. Epistasis in RNA viruses. Science 306:1492-1493.
Netter, H. J., T.-T. Wu, M. Bockol, A. Cywinski, W.-S. Ryu, B. C. Tennant, and J. M. Taylor. 1995. Nucleotide sequence stability of the genome of hepatitis delta virus. J. Virol. 69:1687-1692.
Nie, X., J. Chang, and J. M. Taylor. 2004. Alternative processing of hepatitis delta virus antigenomic RNA transcripts. J. Virol. 78:4517-4524.
Oehlenschlager, F., and M. Eigen. 1997. 30 years later—a new approach to Sol Spiegelman's and Leslie Orgel's in vitro evolutionary studies. Dedicated to Leslie Orgel on the occasion of his 70th birthday. Origins Life Evol. Biosph. 27:437-457.
Polson, A. G., H. L. Ley III, B. L. Bass, and J. L. Casey. 1998. Hepatitis delta virus RNA editing is highly specific for the amber/W site and is suppressed by hepatitis delta antigen. Mol. Cell. Biol. 18:1919-1926.
Radjef, N., E. Gordien, V. Ivaniushina, E. Gault, P. Anais, T. Drugan, J. C. Trinchet, D. Roulot, M. Tamby, M. C. Milinkovitch, and P. Deny. 2004. Molecular phylogenetic analyses indicate a wide and ancient radiation of African hepatitis delta virus, suggesting a deltavirus genus of at least seven major clades. J. Virol. 78:2537-2544.
Sanjuan, R., F. M. Codoner, A. Moya, and S. F. Elena. 2004. Natural selection and the organ-specific differentiation of HIV-1 V3 hypervariable region. Evol. Int. J. Org. Evol. 58:1185-1194.
Sanjuan, R., A. Moya, and S. F. Elena. 2004. The contribution of epistasis to the architecture of fitness in an RNA virus. Proc. Natl. Acad. Sci. USA 101:15376-15379.
Sanjuan, R., A. Moya, and S. F. Elena. 2004. The distribution of fitness effects caused by single-nucleotide substitutions in an RNA virus. Proc. Natl. Acad. Sci. USA 101:8396-8401.
Spiegelman, S., I. Haruna, N. R. Pace, D. R. Mills, D. H. Bishop, J. R. Claybrook, and R. Peterson. 1967. Studies in the replication of viral RNA. J. Cell. Physiol. 70(Suppl.):35-64.
Taylor, J. M. Structure and replication of HDV RNA. In J. L. Casey (ed.), Current topics in microbiology and immunology, in press.
Wang, T. C., and M. Chao. 2005. RNA recombination of hepatitis delta virus in natural mixed-genotype infection and transfected cultured cells. J. Virol. 79:2221-2229.
Wu, J. C., T. Y. Chiang, W. K. Shiue, S. Y. Wang, I. J. Sheen, Y. H. Huang, and W. J. Syu. 1999. Recombination of hepatitis D virus RNA sequences and its implications. Mol. Biol. Evol. 16:1622-1632.
Wu, J. C., S. C. Hsu, S. Y. Wang, Y. H. Huang, I. J. Sheen, H. H. Shih, and W. J. Syu. 2005. "Defective" mutations of hepatitis D viruses in chronic hepatitis D patients. World J. Gastroenterol. 11:1658-1662.
Wu, T.-T., H. J. Netter, D. W. Lazinski, and J. M. Taylor. 1997. Effects of nucleotide changes on the ability of hepatitis delta virus to transcribe, process, and accumulate unit-length, circular RNA. J. Virol. 71:5408-5414.(Jinhong Chang, Severin O.)
ABSTRACT
Previous studies have defined a novel cell culture system in which a modified RNA genome of hepatitis delta virus (HDV) is able to maintain a low level of continuous replication for at least 1 year, using a separate and limited DNA-directed source of mRNA for the essential small delta protein. This mode of replication is analogous to that used by plant viroids. An examination was made of the nucleotide changes that accumulated on the HDV RNA during 1 year of replication. The length of the RNA genome was maintained, except for some single-nucleotide deletions and insertions. There was an abundance of single-nucleotide substitutions, with a 22-fold excess of these being base transitions rather than transversions. Of the detected transitions, at least 70% were consistent with being the consequences of posttranscriptional RNA editing by an adenosine deaminase acting on RNA. The remainder of the changes, including the single-nucleotide insertions and deletions, are likely to be the consequence of misincorporation during transcription. In addition, an intermolecular competition assay was used to show that the majority of the genomes present after 1 year of replication were essentially as competent in replication as the original single HDV RNA sequence that was used to initiate the genome replication. A model is provided to explain how, in this experimental system, the observed single-nucleotide changes were essentially neutral in terms of their effect on the ability of the HDV genome to carry out continued rounds of replication.
INTRODUCTION
Previous studies have considered the nucleotide changes that accumulate during the replication of the approximately 1,700-nucleotide single-stranded circular RNA genome of human hepatitis delta virus (HDV) (1, 12, 15, 22, 25, 28, 37). Some of these changes are considered to arise via misincorporations during RNA-directed RNA transcription, a process that is considered to involve redirection of host DNA-directed RNA polymerase II (34). In contrast, one much-studied nucleotide change that occurs at position 1012, in the middle of the amber termination codon of the open reading frame for the essential small delta protein, Ag-S (23), is not caused by misincorporation. This change occurs during genome replication but is a consequence of posttranscriptional RNA editing by ADAR, an adenosine deaminase acting on RNA (2). This change leads to inactivation of the termination codon and allows the synthesis of a longer protein, the 214-amino-acid large delta protein, Ag-L, that is a dominant negative inhibitor of genome replication (6) and yet is essential for the process of virus assembly which is mediated by the envelope proteins of a helper virus, hepatitis B virus (3). There are other less-site-specific nucleotide changes that have been noted in previous studies, whether during passage of the virus in infected animals or even during replication in cultured cell lines (12, 25, 28). It is no doubt because of the accumulation of changes such as these that the different HDV isolates from around the world can be divided into at least three, and possibly seven, different genotypes (29).
In the present studies, we have made use of a recently described novel system for HDV genome replication in order to follow the accumulation of changes that occur during extended replication of the HDV genome (4). Briefly, the system uses a line of human embryonic kidney cells, 293-HDV, that contain a single copy of a cDNA to express Ag-S, under tetracycline-on control. In addition, the cells have been transfected with an HDV RNA that by the creation of a 2-nucleotide deletion has disrupted the open reading frame for Ag-S (6). This genome, however, can replicate using the unchangeable source of Ag-S from the cDNA. If tetracycline (TET) is added, these cells produce large amounts of Ag-S, leading to extensive HDV replication, cell cycle arrest, and cell death. However, in the absence of TET, there is sufficient Ag-S translated to allow the mutated genome to replicate at the low level of about 1,000 copies per cell, without any indication of cytotoxic effects, for at least 1 year (4).
The strategy of the present study includes the use of reverse transcription-PCR (RT-PCR), cloning, and sequencing to detect and characterize the changes that occurred on the HDV RNA during 1 year of replication. In addition, we used a previously described competition assay to examine the replication competence of the HDV genomes at this time, relative to that of the single HDV RNA sequence that was first used to initiate the genome replication (13).
MATERIALS AND METHODS
Cell culture. The creation and maintenance of low-level HDV genome replication was carried out as previously described (4). Briefly, a line of 293-TREx cells (Invitrogen) was stably transformed with cDNA encoding the HDV small delta antigen under TET control. These cells, designated 293-Ag, were then transfected with an HDV antigenomic RNA containing a 2-nucleotide deletion in the open reading frame for the small delta antigen (6). When these cells were maintained under TET– conditions, a low level of HDV replication (about 1,000 copies per cell) was detected for periods of at least 1 year. These cells were designated 293-HDV.
RNA analysis. Unless specified otherwise, 293-HDV cells were induced with TET (1 μg/ml) for 2 days prior to harvest. Total RNA was extracted using Tri Reagent (Molecular Research Center, Inc.). RT-PCR, cloning, and nucleotide sequencing were carried out as previously described (5). The DNA primers used are described in Fig. 1. HDV RNAs were also examined by Northern analyses using 32P-labeled riboprobes, as previously described (38).
Intermolecular competition assay. This was a modification of a previously described assay that measures the relative replication competence of HDV genomes (13). It was carried out by cotransfection of new cells with a combination of two sources of total RNA. The first was from 293-HDV cells at 1 year. The other was from Huh7 cells at 3 days after transfection with wild-type HDV RNAs that contained a unique marker sequence.
As a preliminary to our competition experiments, we used Northern analyses to normalize the amounts of HDV RNA. Then, approximately equal amounts of the two RNAs were cotransfected into recipient cells. At various times thereafter, total RNA was extracted and assayed to measure the relative replication of these two forms of HDV RNA, using Northern analysis with oligonucleotide probes specific for the marked and unmarked sequences (13).
More specifically, we detected the genomes at 1 year with the following antigenomic oligonucleotide (nucleotides 804 to 784): 5'-CTCGATTCTCTA TCGGAATCT-3'. This oligonucleotide was chosen in a region that was absolutely conserved during 1 year of replication (see Table S1 in the supplemental material). In contrast, the marked wild-type genomes were detected with the oligonucleotide 5'-CTCGATTCTGCTTCAGAATCT-3'. This sequence differs from the wild-type sequences at the nucleotides that are underlined.
RESULTS
Strategy for RT-PCR, cloning, and sequencing. The following strategy was used for the cloning and sequencing of the HDV RNA genomes after 1 year of continuous low-level replication in the 293-HDV cell line. Two days prior to the harvest of total RNA, TET was added to the cells in order to transiently increase the copy number of HDV RNAs and thereby facilitate the subsequent HDV RNA analysis. (As mentioned below, control studies were performed, which showed that such induction for 2 days was not a significant source of nucleotide changes relative to the 1 year of replication in the absence of TET.) Four primer pairs, as indicated in Fig. 1, were used to perform reverse transcription and PCR amplification. These four regions have significant overlap and together represent the entire HDV genome sequence. These PCR products were cloned, and in each case, at least nine clones were subjected to nucleotide sequencing.
Detected sequence changes. Table 1 summarizes the nucleotide sequence changes detected. Figure 2 shows the location of each of these changes relative to the full HDV RNA sequence and the locations of known sequence features of both the genome and the antigenome. Complete information for all changes is presented in Table S1 of the supplemental material.
For 26,761 total nucleotides sequenced, we detected 313 changes at 121 different sites relative to the HDV sequence that was first used to initiate replication in these cells. This represents about 2.1% changes per nucleotide per year. This value seems high relative to the range of 0.001 to 0.34% for 50 different RNA viruses (17); however, as discussed below, we estimate that around 70% of our detected changes may have arisen not as cotranscriptional misincorporations but as posttranscriptional RNA editing events.
There were no major changes in the length of the HDV genome. The only length changes detected were of relatively rare single-nucleotide insertions and deletions. These, representing only 7% of all changes, occurred at several different locations. (In a previous study of passaged HDV, such changes were not detected, that is, represented <2.5% of all changes [25].)
The majority of the detected changes (93%) were single-nucleotide substitutions. Of these, the majority were base transitions rather than transversions. Furthermore, not all possible base transitions were observed with comparable frequencies. The CU and GA changes combined were less than 10%. In contrast, 90% of the changes were consistent with either a UC or an AG.
Two sites underwent nucleotide changes in 100% of clones sequenced. The first, located at position 1375, has been reported in one previous study and is probably an error due to a misincorporation during RNA-directed RNA synthesis (12). Now, from studies by us and by others it is established that the second change, at position 1012, arises via posttranscriptional RNA editing by an adenosine deaminase acting on RNA, an ADAR (2, 23). This editing converts AI on the antigenome and, with further replication, changes the genome from UC and the antigenome from AG. Knowing this, and knowing that other sites of putative ADAR editing have been shown on HDV RNAs (25), we would interpret that most of the UC or AG changes detected on the genomic sequence are the consequences of ADAR editing on either genomic or antigenomic RNAs. From this assumption, we deduce that about 70% (79.6 – 9.2%) of all of the detected changes could have arisen via ADAR editing, with 18% (9.2 + 9.2%) being base transitions due to misincorporation. Further support for this interpretation was gained by an examination of the 5' neighbors for these putative ADAR editing sites. That is, the frequencies for the observed 5'-neighboring nucleotides were very similar to those of the naturally occurring editing sites on RNAs with secondary structure (18) (see Table S2 in the supplemental material for more details.)
If as many as 70% of the changes could be due to ADAR editing and 7% are single-nucleotide insertions and deletions, the remaining 23% are probably the consequence of misincorporations during transcription. Three different changes were detected at position 553; this might suggest some template structure enhanced transcriptional errors. Similarly, at position 611 there was a frequent deletion of one nucleotide.
However, we do have to allow that some fraction of these detected changes may have arisen as a consequence of our in vitro analysis. We estimate the combined error rate for RT-PCR, cloning, and associated nucleotide sequencing to be <0.2% (12). With this in mind, in Fig. 2 we indicated changes that were detected only once. It can be noted that of the changes detected only once, 78% were UC or AG, which could be explained as ADAR editing. (See Table S1 in the supplemental material for more details.)
We carried out certain controls to determine whether the changes were accumulating with the extent of HDV replication. One control was to examine the replicating HDV RNA sequences at the earlier time of 16 weeks. This was done for the region flanked by the primer pair A, shown in Fig. 1. We definitely found changes at this time, especially at position 1012, and yet the extent was noticeably less than that discovered at 1 year. Thus, at 1 year we were detecting changes that accumulated with time of replication (data not shown). Another control was to determine whether significant changes arose during the brief 2-day period of TET induction that was used to amplify the intracellular HDV RNA levels, thus facilitating the RT-PCR strategy. We therefore compared the sequences of clones made from the region flanked by primer pair A (Fig. 1) for RNA from uninduced cells at 1 year relative to those after the 2 days of TET induction. Both contained the same spectrum of nucleotide changes (data not shown).
We also considered the detected sequence changes in terms of frequency and location relative to the sequence motifs that can be defined for the HDV genome and antigenome. Figure 2 is such a representation. It can be seen, as mentioned above, that the most frequently detected changes were at position 1012, within the termination codon for Ag-S, and at position 1375.
Several other sites were also abundant. Most striking was the cluster of changes in the region at and surrounding the poly(A) signal, AAUAAA. The changes in this region are shown in greater detail in Fig. 3, representing a total of 17 independent sequences. Strikingly, this cluster of changes (like that at position 1012) is consistent with ADAR editing, either on the antigenome or genome. To test whether these changes had an impact on HDV poly(A) processing, we used Northern analyses. As shown in Fig. 4, we observed not one but two species of HDV mRNA derived from the replication of these genomes. Relative to controls, we deduce that one species was as normally found during HDV genome replication while the second migrated somewhat more quickly. It is possible that the faster species contained a shorter poly(A) tail. Maybe the poly(A) signal changes on the genomes at 1 year interfered with addition of the poly(A) tail, for example, as an intrinsic fault during HDV poly(A) processing and/or an inability to maintain a typical full-length poly(A) tail.
In considering these data, it is important to keep in mind that in our experimental system, translation of Ag-S from the mRNA produced during HDV genome replication is no longer needed. Nevertheless, our data show that even after 1 year of replication and all the accumulated changes, one or more mRNA species continue to be processed. Since some form of poly(A) processing was still maintained, we are tempted to conclude that this processing, independent of the need for Ag, is somehow an essential part of the replication scheme.
Finally, we noted that some regions of the HDV sequence had relatively few, if any, changes. These regions include the two ribozyme domains which are needed for HDV RNA processing and known to be essential for HDV replication (16). Totally conserved was the end of the rod-like structure that we refer to as the top. This region is considered to include the site for the initiation of the HDV mRNA (10, 11). Another relatively conserved region is between positions 219 to 478; we currently have no explanation for this conservation.
Replication competence. Figure 5A summarizes the result of an intermolecular competition assay to determine the replication competence of the genomes at 1 year relative to that of the RNA used to initiate replication in the first place. This strategy was carried out largely as previously described (13). The assay involves cotransfection of recipient TET-induced 293-Ag cells with the two sources of HDV RNA, and at various times thereafter, total RNA is extracted and assayed by Northern analysis with oligonucleotide probes that can specifically detect the genomes at 1 year and the wild-type genomes. Any change in the ratio of these genomes is considered a measure of the relative ability to initiate and continue HDV replication. In the cotransfection, each cell receives multiple copies of each type of genome and they share the available Ag.
Three experiments were performed with the data for the average values and the range of these values as represented in Fig. 5A. As shown, the ratio of genomes at 1 year to wild-type genomes did not change significantly between day 0 and day 1. This means that the HDV RNAs at 1 year were just as competent as the wild type in initiating HDV genome replication. They were neither more nor less competent. Furthermore, this ratio did not change for times after 1 day. Therefore, the genomes at 1 year were not at any significant genetic advantage or disadvantage, since either of these could lead to a time-dependent change in the ratio.
We next asked whether the genomes at 1 year, even though they were fully replication competent, had actually been selected for a replication competence that was unique to our competition assay system that uses the 293-Ag cells during TET induction. We tested the ability of HDV at 1 year to compete with wild-type HDV RNA in 293 cells when the sources and/or amounts of available Ag were varied. First, we tested the competition in 293-Ag cells when there was no TET induction (Fig. 5B). Also, we tested competition in 293 cells when the only source of Ag was cotransfection of the relevant mRNA as first transcribed in vitro. In both cases, we found the genomes at 1 year to have replication competence comparable to that of the wild type (Fig. 5B).
However, a different answer was obtained when we tested replication competence in cells other than those related to 293 cells, such as Huh7 and HeLa cells. For these competition assays, the HDV RNAs were again cotransfected along with in vitro-transcribed mRNA to express the essential Ag-S. In both cases, there was a sixfold drop in the ratio (Fig. 5B) and these reduced ratios did not decrease further with time after transfection (data not shown). One interpretation is that, when assayed under these conditions, only 15% of the genomes at 1 year were as competitive as the wild type for initiation of HDV replication in these cells and that these genomes that did initiate replication were not at any genetic disadvantage relative to the wild type.
It was important to determine whether those genomes at 1 year that could replicate when passaged into fresh 293 or Huh7 cells in the absence of competitor wild-type virus maintained the many nucleotide sequence changes. Therefore, at 4 days after transfection, the total RNA was extracted and the HDV sequence flanked by the primer pair A in Fig. 1 was subjected to RT-PCR, cloning, and sequencing. We observed that the sequences obtained still contained many single-nucleotide changes, comparable to the genomes at 1 year that were used to initiate this additional round of replication (data not shown). Thus, while we do favor that there was a selection associated with replication in different host cells, the data argue against an obvious selection for a minor population of HDV genomes, for example, with relatively fewer sequence changes.
DISCUSSION
The present experiments have provided new and important information on the evolution of replicable RNA molecules over time. Previously, others have tried to follow the evolution of RNA molecules that can be replicated in the test tube. Two important examples of this are studies of the phage Q? mini-replicon (27, 33) and the short RNAs replicated by a phage RNA polymerase (19, 20). In our studies, the HDV RNA genome was replicated in a cell line that had been transfected with RNA 1 year earlier. There was no cell-to-cell spread of these HDV genomes other than at what must be about 365 consecutive cell divisions. Throughout all of this, the HDV replication was totally dependent upon an integrated cDNA source of fully functional, unchanging Ag-S. In this way, the cells maintained about 1,000 copies of HDV genomic RNA per cell.
We consider that at least two negative selective pressures acted against the survival of replicable HDV RNA in our experimental system. The first was the accumulation of errors during RNA-directed RNA transcription. The second was consistent with posttranscriptional RNA editing by ADAR activity. Such editing altered virtually all of the RNAs at position 1012, but there were in addition many single-nucleotide changes that were of the specific form consistent with ADAR editing of either genomic or antigenomic RNAs. From Table 1, we might thus deduce that among the RNA genomes that accumulated, ADAR editing was responsible for more changes that were tolerated than was nucleotide misincorporation. Other than the present study, there are two published reports where HDV nucleotide sequences changes were followed after beginning replication with a single HDV sequence (25, 28). Both studies examined only about 350 nucleotides of the HDV genome. In one study, HDV was serially passaged by infection of woodchucks, using woodchuck hepatitis virus as the helper virus. In the other study, the sequence changes were detected following a transfection with cDNA. In both studies, like in the present study, at least 70% of the single-nucleotide changes observed could have been caused by ADAR editing.
While nucleotide misincorporation during transcription and ADAR editing acting posttranscription might have been negative pressures acting on HDV genome replication, as now explained, we consider that there was also a positive selective pressure to maintain genome replication. In our experimental system, each cell contains about 1,000 copies of HDV genomic RNA. As each cell divides, another 1,000 copies need to be transcribed, processed, and accumulated. If a fraction of the original 1,000 copies had been compromised by the above-mentioned negative pressures and become less replication competent, then most of the new 1,000 copies would be made from the noncompromised genomes. This purging effect would be achieved with every cell division over the 1-year period of study. Thus, without the wisdom of hindsight, we should have expected that after 1 year, the surviving genomes would all be just as replication competent as the original wild-type sequence. In addition, those that had undergone any nucleotide changes would be such that the changes were totally neutral and not even detectably compromised in replication competence relative to the wild type. In retrospect, we also realize that if any of the changed genomes were able to replicate not less but more efficiently, this in turn might also be selected against. After all, as previously reported, enhanced replication of HDV RNA in this cell system can lead to cell cycle arrest followed by cell detachment and death (4).
Since all of the genomes sequenced at 1 year contained about 20 nucleotide changes, many of which were at diverse locations, we conclude that these were changes that did not detectably interfere with replication competence. That is, these genomes contained all of the necessary cis-acting sequences and structural features essential for uncompromised replication (Fig. 2).
The nature of our experimental system is that these conserved features do not include maintaining the ability to make mRNA that can be translated to make the essential Ag-S. Thus, in our experimental system HDV genome replication is mimicking that replication which is achieved by viroids. These are infectious subviral agents of plants with small, single-stranded, circular RNA genomes whose replication is totally dependent on host-encoded proteins (8).
One prediction of the above interpretation of a positive selective pressure is that if a change arises which compromises genome replication, there may not be an opportunity for the appearance of a second-site change that might rescue the genome. If the only changes that can accumulate with time are those that, per se, are totally neutral, then to explain the observed total of about 20 changes per genome, we predict that these must have accumulated over time from a series of single-nucleotide changes, each of which was totally neutral.
For many RNA viruses, there has been extensive thinking and some controversy, in terms of the concept of quasi-species, about which heterogeneous populations of viral genomes can provide mutual support to achieve a level of replication (7). Maybe this concept is not applicable to HDV replication as it is examined in our studies. This is because none of the genomes that are replicating can make a functional Ag-S. All of the genomes are dependent upon a separate source of this Ag-S. However, if the HDV replication were not supported by an external source then the quasi-species model could apply to some extent. That is, if some of the viral genomes were providing a source of functional Ag-S, they would be able to support, in trans, the replication of other genomes that were unable to make any Ag or that made a Ag-S that was not fully functional. Of course, if the supported genomes were making a significant amount of a form of Ag-S that interfered with HDV replication (like Ag-L), then they may compromise the coreplication. It should also be noted that some aspects of the quasi-species models demand that the replicating genomes can undergo significant levels of intermolecular recombination. For HDV, this has been claimed (35, 36), but the issue is still controversial (14, 34).
What about epistasis? Recent studies have addressed the issue of whether this can contribute to the fitness of an RNA virus (9, 24, 30-32). The discussion has included concepts of both positive and negative epistasis. It might be considered that our studies do not impact this question because we do not have clones for individual full-length HDV RNAs. One would have to obtain such clones and test them individually for their ability to initiate HDV replication. However, our evidence that the major fraction of the genomes at 1 year are as replication competent as the wild type, together with the observation that the majority have also undergone sequence changes, supports the interpretation that such changes have not conspired to produce combinations of positive and negative epistatic effects. At the same time, we cannot exclude that deliberately introduced changes might not achieve such effects.
Finally, we come to the question of what happens in terms of replication fitness when the HDV RNAs at 1 year are transfected to initiate replication in cells other than the 293 cells in which they were maintained for 1 year. We found from our competition assays that in other human cell lines, whether of liver or nonliver origin, only about 15% of the genomes had a replication competence that was equivalent to that of wild-type genomes (Fig. 5B). What then was the disadvantage for the 85% that were unable to initiate replication? Maybe these altered genomes were of a secondary structure that could not be adequately protected by Ag-S when expressed in Huh7 and HeLa cells.
In summary, the present study has been able to exploit a new model of HDV replication to examine the long-term consequences of such replication on the integrity of the HDV genomes. The findings have significant implications in terms of (i) the conservation of essential cis-acting sequence features, (ii) the fidelity of RNA-directed RNA synthesis by a host polymerase, and (iii) the impact of ADAR editing.
ACKNOWLEDGMENTS
This work was supported by grants AI-26522 and CA-06927 from the NIH and by an appropriation from the Commonwealth of Pennsylvania.
We thank Anita Cywinski and the Fox Chase DNA Sequencing Facility (Randy Hardy, manager). Constructive comments on the manuscript were given by Sven Behrens and Richard Katz. We thank Raul Andino for pointing out the need to test replication fitness in multiple cell types.
Supplemental material for this article may be found at http://jvi.asm.org/.
REFERENCES
Casey, J. L., T. L. Brown, E. J. Colan, F. S. Wignall, and J. L. Gerin. 1993. A genotype of hepatitis D virus that occurs in northern South America. Proc. Natl. Acad. Sci. USA 90:9016-9020.
Casey, J. L., and J. L. Gerin. 1995. Hepatitis D virus RNA editing: specific modification of adenosine in the antigenomic RNA. J. Virol. 69:7593-7700.
Chang, F. L., P. J. Chen, S. J. Tu, M. N. Chiu, C. J. Wang, and D. S. Chen. 1991. The large form of hepatitis antigen is crucial for the assembly of hepatitis virus. Proc. Natl. Acad. Sci. USA 88:8490-8494.
Chang, J., S. O. Gudima, C. Tarn, X. Nie, and J. M. Taylor. 2005. Development of a novel system to study hepatitis delta virus genome replication. J. Virol. 79:8182-8188.
Chang, J., and J. Taylor. 2002. In vivo RNA-directed transcription, with template switching, by a mammalian RNA polymerase. EMBO J. 21: 157-164.
Chao, M., S.-Y. Hsieh, and J. Taylor. 1990. Role of two forms of the hepatitis delta virus antigen: evidence for a mechanism of self-limiting genome replication. J. Virol. 64:5066-5069.
Domingo, E. 2000. Viruses at the edge of adaptation. Virology 270:251-253.
Flores, R., S. Delgado, M. E. Gas, A. Carbonell, D. Molina, S. Gago, and M. De la Pena. 2004. Viroids: the minimal non-coding RNAs with autonomous replication. FEBS Lett. 567:42-48.
Froissart, R., D. Roze, M. Uzest, L. Galibert, S. Blanc, and Y. Michalakis. 2005. Recombination every day: abundant recombination in a virus during a single multi-cellular host infection. PLoS Biol. 3:e89.
Gudima, S., K. Dingle, T.-T. Wu, G. Moraleda, and J. Taylor. 1999. Characterization of the 5' ends for polyadenylated RNAs synthesized during the replication of hepatitis delta virus. J. Virol. 73:6533-6539.
Gudima, S., S.-Y. Wu, C.-M. Chiang, G. Moraleda, and J. Taylor. 2000. Origin of the hepatitis delta virus mRNA. J. Virol. 74:7204-7210.
Gudima, S. O., J. Chang, G. Moraleda, A. Azvolinsky, and J. Taylor. 2002. Parameters of human hepatitis delta virus replication: the quantity, quality, and intracellular distribution of viral proteins and RNA. J. Virol. 76: 3709-3719.
Gudima, S. O., J. Chang, and J. M. Taylor. 2004. Features affecting the ability of hepatitis delta virus RNAs to initiate RNA-directed RNA synthesis. J. Virol. 78:5737-5744.
Gudima, S. O., J. Chang, and J. M. Taylor. 2005. Reconstitution in cultured cells of replicating HDV RNA from pairs of less than full-length RNAs. RNA 11:90-98.
Imazeki, F., M. Omata, and M. Ohto. 1990. Heterogeneity and evolution rates of delta virus RNA sequences. J. Virol. 64:5594-5599.
Jeng, K.-S., A. Daniel, and M. M. C. Lai. 1996. A pseudoknot structure is active in vivo and required for hepatitis delta virus RNA replication. J. Virol. 70:2403-2410.
Jenkins, G. M., A. Rambaut, O. G. Pybus, and E. C. Holmes. 2002. Rates of molecular evolution in RNA viruses: a quantitative phylogenetic analysis. J. Mol. Evol. 54:156-165.
Knight, S. W., and B. L. Bass. 2002. The role of RNA editing by ADARs in RNAi. Mol. Cell 10:809-817.
Konarska, M. M., and P. A. Sharp. 1989. Replication of RNA by DNA-dependent RNA polymerase of phage T7. Cell 57:423-431.
Konarska, M. M., and P. A. Sharp. 1990. Structure of RNAs replicated by the DNA-dependent T7 RNA polymerase. Cell 63:609-618.
Kuo, M. Y. P., J. Goldberg, L. Coates, W. Mason, J. Gerin, and J. Taylor. 1988. Molecular cloning of hepatitis delta virus RNA from an infected woodchuck liver: sequence, structure, and applications. J. Virol. 62: 1855-1861.
Lee, C. M., F. Y. Bih, Y. C. Chao, S. Govindarajan, and M. M. C. Lai. 1992. Evolution of hepatitis delta virus RNA during chronic infection. J. Virol. 188:265-273.
Luo, G., M. Chao, S.-Y. Hsieh, C. Sureau, K. Nishikura, and J. Taylor. 1990. A specific base transition occurs on replicating hepatitis delta virus RNA. J. Virol. 64:1021-1027.
Michalakis, Y., and D. Roze. 2004. Evolution. Epistasis in RNA viruses. Science 306:1492-1493.
Netter, H. J., T.-T. Wu, M. Bockol, A. Cywinski, W.-S. Ryu, B. C. Tennant, and J. M. Taylor. 1995. Nucleotide sequence stability of the genome of hepatitis delta virus. J. Virol. 69:1687-1692.
Nie, X., J. Chang, and J. M. Taylor. 2004. Alternative processing of hepatitis delta virus antigenomic RNA transcripts. J. Virol. 78:4517-4524.
Oehlenschlager, F., and M. Eigen. 1997. 30 years later—a new approach to Sol Spiegelman's and Leslie Orgel's in vitro evolutionary studies. Dedicated to Leslie Orgel on the occasion of his 70th birthday. Origins Life Evol. Biosph. 27:437-457.
Polson, A. G., H. L. Ley III, B. L. Bass, and J. L. Casey. 1998. Hepatitis delta virus RNA editing is highly specific for the amber/W site and is suppressed by hepatitis delta antigen. Mol. Cell. Biol. 18:1919-1926.
Radjef, N., E. Gordien, V. Ivaniushina, E. Gault, P. Anais, T. Drugan, J. C. Trinchet, D. Roulot, M. Tamby, M. C. Milinkovitch, and P. Deny. 2004. Molecular phylogenetic analyses indicate a wide and ancient radiation of African hepatitis delta virus, suggesting a deltavirus genus of at least seven major clades. J. Virol. 78:2537-2544.
Sanjuan, R., F. M. Codoner, A. Moya, and S. F. Elena. 2004. Natural selection and the organ-specific differentiation of HIV-1 V3 hypervariable region. Evol. Int. J. Org. Evol. 58:1185-1194.
Sanjuan, R., A. Moya, and S. F. Elena. 2004. The contribution of epistasis to the architecture of fitness in an RNA virus. Proc. Natl. Acad. Sci. USA 101:15376-15379.
Sanjuan, R., A. Moya, and S. F. Elena. 2004. The distribution of fitness effects caused by single-nucleotide substitutions in an RNA virus. Proc. Natl. Acad. Sci. USA 101:8396-8401.
Spiegelman, S., I. Haruna, N. R. Pace, D. R. Mills, D. H. Bishop, J. R. Claybrook, and R. Peterson. 1967. Studies in the replication of viral RNA. J. Cell. Physiol. 70(Suppl.):35-64.
Taylor, J. M. Structure and replication of HDV RNA. In J. L. Casey (ed.), Current topics in microbiology and immunology, in press.
Wang, T. C., and M. Chao. 2005. RNA recombination of hepatitis delta virus in natural mixed-genotype infection and transfected cultured cells. J. Virol. 79:2221-2229.
Wu, J. C., T. Y. Chiang, W. K. Shiue, S. Y. Wang, I. J. Sheen, Y. H. Huang, and W. J. Syu. 1999. Recombination of hepatitis D virus RNA sequences and its implications. Mol. Biol. Evol. 16:1622-1632.
Wu, J. C., S. C. Hsu, S. Y. Wang, Y. H. Huang, I. J. Sheen, H. H. Shih, and W. J. Syu. 2005. "Defective" mutations of hepatitis D viruses in chronic hepatitis D patients. World J. Gastroenterol. 11:1658-1662.
Wu, T.-T., H. J. Netter, D. W. Lazinski, and J. M. Taylor. 1997. Effects of nucleotide changes on the ability of hepatitis delta virus to transcribe, process, and accumulate unit-length, circular RNA. J. Virol. 71:5408-5414.(Jinhong Chang, Severin O.)