INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

For several years only three RNAs have been the focus of hepatitis delta virus (HDV) research and models to explain replication. These are (i) the unit-length 1,679-nucleotide (nt) circular RNA genome, (ii) its exact complement, the circular antigenome, and (iii) relatively small amounts of a subgenomic polyadenylated RNA that has been assumed to be the mRNA for the translation of the delta protein (2, 5-7, 12, 14).

In terms of the genomic RNA, while most of the species are circular in conformation, we and others have noted (but then largely ignored) the presence of unit-length RNAs that are linear (2). These linear forms have been assumed to be molecules (i) unclosed or reopened at nt 686, the cleavage site of the genomic ribozyme, or (ii) formerly circular molecules that have been randomly nicked, just once, by a process that occurred either within the cell or during the extraction procedure. We present here data to show that both of these assumptions are largely wrong, in that a major fraction of the linear genomic RNAs are open at a single specific site (nt 1212). Moreover, this opening is not directed by the genomic ribozyme.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Infected woodchucks. Two animals chronically infected with woodchuck hepatitis B virus were superinfected with HDV. At the peak of the HDV infection (21 to 25 days), the animals were sacrificed and the liver tissue was collected and stored at 80°C (17).

Improved method of RNA extraction. In previous studies we have extracted RNA from tissue by a guanidine isothiocyanate procedure that involved the addition of cesium chloride followed by centrifugation to equilibrium in order to obtain a band of RNA (3). As an improvement to this procedure, we added the frozen tissue directly into 10 volumes of Tri-Reagent (Molecular Research Center) followed by immediate processing (except as specified for Fig. 6) in a Brinkmann homogenizer. We then followed the manufacturer's instructions to collect the RNA free of DNA and protein.

In vitro activation of genomic ribozyme. To achieve ribozyme cleavage of the genomic HDV RNA extracted from the infected livers, we used a modification of a previously described procedure (11). Briefly, we first synthesized in vitro an antigenomic RNA corresponding to the SalI (nt 962)-to-XbaI (nt 781) region of the HDV RNA. This RNA was then hybridized to an aliquot of the liver RNA. Previous studies show that this hybridization allows the genomic ribozyme to fold into an active state. In a subsequent heating in the presence of magnesium ions, the genomic RNA is cleaved by the activated ribozyme.

Northern analyses. HDV RNA species were detected by Northern analysis subsequent to glyoxalation and electrophoresis. Using gels of 3% agarose and narrow wells, we were able to achieve separation of the linear and circular forms of HDV RNA (2). Radioactive RNA probes were to the entire HDV sequence or were region specific. Subsequent detection and quantitation were obtained using a Fuji Bio-Imaging system.

Primer extension analyses. To detect cleavage at the ribozyme site (nt 686), we used antigenomic primer nt 766-741 (5'-CCATTCGCCATTACCGAGGGGACGGT); to detect the opening site, ultimately proven to be nt 1212, we used antigenomic primer nt 1301-1267 (5'-CAGGATCACCGACGAAGGAAGGCCCTCGAGAACAA). These primers were 5' labeled using [-³²P]ATP (7,000 Ci/mmol; ICN), hybridized to the liver RNA, and then extended with reverse transcriptase (Superscript II; Life Technologies). When needed, a dideoxynucleotide sequencing ladder was obtained using the same labeled primer, and pDL444 (10) as the DNA template, in a cycle-sequencing reaction according to the instructions of the manufacturer (Promega). Products were analyzed on a sequencing gel of 10% polyacrylamide in the presence of 8 M urea.

Modified RLM-RACE to detect and characterize 5' ends. Most aspects of our strategy are represented in Fig. 3. Initially we found that unit-length linear genomic RNA, which can fold into the rod-like structure (9), was not a good substrate for RNA ligase-mediated rapid amplification of cDNA ends (RLM-RACE). We solved this by precleaving the RNA at a second site, to produce two separable fragments with exposed ends. To achieve this, 40 µg of total liver RNA from an infected woodchuck was annealed to antigenomic oligonucleotide nt 325-302 (5'-CGCTGAAGGGGTCCTCTGGAGGTG) and then treated with RNase H (Life Technologies). The reaction product was extracted with phenol, collected by ethanol precipitation, and resuspended at a concentration of about 1 mg/ml in sodium acetate (0.1 M, pH 5.0). To block the 3'-OH ends, a 100-fold molar excess of sodium metaperiodate (Fisher Chemical) was added, and the sample was then incubated for 45 min at room temperature in the dark (8). Next the HDV-specific genomic RNA was selected by annealing to the biotin-labeled antigenomic oligonucleotide nt 286-252 (5'-biotin-TCCGAGTGGATTCCTCCCTCTGAGTGCTACTCAAC), followed by an affinity selection to 0.5 mg of streptavidin-coated superparamagnetic beads (Dynal). The bound HDV-specific RNA was then divided into four aliquots for the following specific treatments, based on modifications of an RLM-RACE kit (Ambion) to detect and characterize four possible 5' ends at the novel opening site. (i) To detect what might be a 5' cap, an aliquot was treated with calf intestinal alkaline phosphatase (CIP) to remove all existing 5'-phosphate groups and then treated with tobacco acid pyrophosphatase (TAP) to remove the cap and create a new 5'-monophosphate. (ii) To detect a 5'-triphosphate, an aliquot was treated with TAP to create a 5'-monophosphate (21). (iii) To detect a 5'-OH, an aliquot was treated with T4 polynucleotide kinase to create a 5'-monophosphate. (iv) Finally, to detect a 5'-monophosphate, no pretreatment was needed. Subsequent steps were as described for the kit. The four aliquots, still immobilized to the beads, were subjected to RNA ligation in the presence of an RNA adapter. Then reverse transcription was carried out, with the biotin-labeled HDV-specific oligonucleotide acting as primer. After this reaction, the beads were treated with alkali to remove the RNA, leaving the DNA product bound to the beads. Nested PCR was carried out using pairs of HDV- and adapter-specific primers. The outer HDV-specific primer was antigenomic oligonucleotide nt 1630-1601 (5'-AAGAGTACTGAGGACGGCCGCCTCTAGCCG), and the inner HDV primer was the same as that used in primer extension (nt 1301-1267). Outer and inner adapter-specific primers were provided by the RLM-RACE kit. The products of the second PCR were cloned using a TOPO TA Cloning kit (Invitrogen). Positive clones were selected by hybridization with 5'-labeled antigenomic oligonucleotide nt 1257-1232 (5'-GCTATCGGCGGGAGGCAAGAACCTCA) and subjected to automated nucleotide sequencing to determine the junction between the 3' end of the adapter and the 5' end of the HDV RNA.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Detection of a novel opening on genomic RNA by Northern analyses. As part of studies of the HDV-specific RNAs in the livers of infected woodchucks, we used an extraction procedure that produced RNAs of higher quality. As a rigorous criterion for RNA quality, we used special gel electrophoretic procedures (2) to separate the circular and linear forms of the unit-length HDV antigenomic and genomic RNAs (Fig. 1A, lanes 1 and 2, respectively). These RNAs were then detected by Northern analysis, and the radioactivity was quantitated (Fig. 1B, lanes 1 and 2, respectively). We found that 96% of the antigenomic RNA but only about 50% of the genomic RNA was circular. Several possibilities were considered to account for this high proportion of linear genomic RNAs. (i) The unit-length linear forms produced by ribozyme cleavage had not been ligated. (ii) The RNAs had been ligated to form circles but subsequently had reopened at the ribozyme site. (iii) The circles had been opened by single randomly located endonucleolytic nicks. (iv) A final possibility, the one that ultimately explained the majority of the openings, was that the circles were already open at one specific site, different from the ribozyme site at nt 686. 

View larger version (41K): [in this window] [in a new window]   FIG. 1.   Northern analysis of circular and linear conformations of antigenomic and genomic RNAs detected in infected liver. Extracted RNA was analyzed before (lanes 1 and 2) or after (lanes 3) activation of the genomic ribozyme by means of an antiattenuator sequence as described in the text. RNA samples were then glyoxalated and electrophoresed into a gel of 3% agarose. Northern analysis was used to detect antigenomic (lane 1) or genomic (lanes 2 and 3) RNA. (A) Radioactivity as detected with a Fuji Bio-Imager; (B) corresponding profiles. The species indicated as circle, linear, and fragments 1 and 2 are as described in the text.

Location of the opening site as determined by primer extension. To determine the precise location of the opening(s) around nt 1200, we used primer extension along with a dideoxynucleotide sequencing ladder to determine the location on the nucleotide sequence. As shown in Fig. 2, we found that the majority of the primer extension products end at nt 1210, with a minor amount at nt 1211. From this we might infer that the majority of the openings occur between nt 1209 and 1210, with fewer between 1210 and 1211. 

View larger version (123K): [in this window] [in a new window]   FIG. 2.   Primer extension and dideoxynucleotide sequencing to determine the location of the novel opening site on genomic RNA. The RNA used was total RNA extracted from the liver of an infected woodchuck. The first four lanes represent the sequencing ladder for C, T, A, and G, respectively; the final lane shows the primer extension product. A small amount of the latter was admixed with the C, T, A, and G lanes, as an internal control to facilitate correct alignment of the 5' ends. At the right is shown the alignment of this with the known sequence determined by Kuo et al. (9).

Location and characterization of the opening site by RLM-RACE. We next used a separate approach, a form of RLM-RACE, to independently determine the location of the opening site and also characterize of the 5' end created by that opening. Initially we applied this method (described in Materials and Methods) directly to the total RNA isolated from the liver of an infected woodchuck. Such studies were unsuccessful until we added the three steps represented in Fig. 3A. The first was to further cleave the unit-length linear genomic RNAs that were opened at around nt 1210. The rationale was to stop these RNAs from making their 5' ends inaccessible, through folding into the known rod-like structure (9). This further cleavage was achieved at a distant site by means of hybridization of a specific oligonucleotide followed by digestion with RNase H. In step 2 we hybridized to the RNA another specific oligonucleotide, one that contained a 5'-biotin group. Then, in step 3, we used affinity selection to streptavidin-coated superparamagnetic beads to immobilize the genomic fragment whose 5' end we wished to locate and characterize.

View larger version (38K): [in this window] [in a new window]   FIG. 3.   Application of RLM-RACE to determine the location and character of the 5' end at the opening site on genomic RNA. (A) Representation of the three steps applied to the RNA from the liver of an HDV-infected woodchuck. The opening on the genomic HDV at around nt 1210 is indicated by two small rectangles, and the uncleaved genomic ribozyme site is indicated by a solid circle. As described in the text, step 1 is the hybridization at a distant location of a specific oligonucleotide (open object), after which this region is digested with RNase H. In step 2, a biotinylated oligonucleotide (open object with solid diamond) is hybridized so as to allow in step 3, affinity binding to streptavidin-coated superparamagnetic beads (shaded circle). (B) The four different pretreatments, a to d, then given to the 5'-ends of the immobilized genomic RNA fragments. (C) Agarose gel analysis of the PCR products obtained after RLM-RACE of the four treated RNAs (lanes a to d). Also shown is a negative control (lane n) in the absence of added template. A ladder of DNA size markers is shown in lane m, with the position of the expected PCR product, 128 bp, indicated at the right. (D) Nucleotide sequence of the adapter-target junction for the clones obtained from the PCR product of panel C, lane c. Also shown are the relevant regions of the HDV genomic RNA sequence, with numbering as in reference 9, and the sequence to the 3' end of the RNA adapter (as provided by the kit manufacturer, Ambion). The open arrow points to the location of the observed adapter-target junction, indicating that the 5' end of the genomic RNA corresponded to opening between nt 1211 and 1212.

Is the opening site created by the genomic ribozyme? The above studies indicated that the opening at nt 1212 produced a 5'-OH. Since we know that the HDV ribozymes produce a 5'-OH and a 2',3'-cyclic monophosphate (20), we were obliged to test whether the opening at nt 1212 was a consequence of ribozyme cleavage. To do this we examined liver RNAs both before and after the ribozyme cleavage step described above, using primer extension assays for the openings at both nt 1212 and 686. The locations of the two primers in relation to the two sites are diagramed in Fig. 4A, with typical results shown in Fig. 4B and C. We observed that opening at nt 1212 had occurred prior to the activation of the genomic ribozyme and was not increased by such activation (Fig. 4B). In contrast, the action of the genomic ribozyme at nt 686 was enhanced around 33-fold by the activation (Fig. 4C). We can thus deduce that (i) the ribozyme could be activated to cut at nt 686, (ii) under such conditions it did not act at nt 1212, and finally (iii) it is unlikely that the prior opening at nt 1212 was in any way directed by that ribozyme.

View larger version (31K): [in this window] [in a new window]   FIG. 4.   Examination of novel opening and ribozyme cleavage on genomic RNAs using primer extension analysis. (A) Primers used to detect opening of HDV genomic RNAs at both nt 1211/1212 and nt 685/686, along with the expected sizes of the primer extension products. (B and C) Corresponding primer extensions to detect 1211/1212 and 685/686, respectively, for liver RNA both before () and after (+) the activation of the genomic ribozyme.

View larger version (16K): [in this window] [in a new window]   FIG. 5.   Diagram of circular and linear unit-length genomic RNAs detected in infected liver, both before and after in vitro activation of the genomic ribozyme. The circle represents the genomic ribozyme cleavage site at 685/686, and the square represents the novel opening site that is located at 1211/1212. Our studies show that most of the unit-length linear genomic RNA is not like species (ii) but rather like species (iii).

Did the opening actually occur in vivo? The above studies show that the opening of genomic RNA at nt 1212 occurs under conditions where 96% of the antigenomic RNA is uncleaved. Even then, this still leaves unanswered the question of whether the opening occurs (i) in vivo, in the liver, (ii) during the process of liver freezing and storage at 80°C, or (iii) during the extraction of RNA from the frozen liver. As one approach we isolated RNA from aliquots of frozen liver that were allowed to undergo some incubation at room temperature prior to extraction for periods of up to 30 min. The aim was to determine whether such conditions would increase the opening at nt 1212. As shown in Fig. 6, we analyzed the extracted RNAs by Northern analysis and primer extension. The Northern analyses showed that as the incubation time was increased, relatively low amounts of subgenomic-sized species appeared. However, the amount of intact circular RNA did not change significantly (Fig. 6A), and when the same RNA samples were assayed by primer extension to detect the specific 90-nt product corresponding to the site at nt 1212, we did not detect a time-dependent increase in the amount of this product (Fig. 6B). We interpret this as evidence that the site at nt 1212 was opened very early (most likely in vivo) and in a manner independent of RNase activity during incubation prior to extraction. It is also worth noting that in our primer extension analyses of RNA from infected liver, we detected lower amounts of several discrete species other than the 90-nt product. We interpret these extra species as the consequence of additional openings on the genomic RNA. We excluded the possibility that these were the consequences of pauses in reverse transcription since such species were not detected when primer extension was carried out on intact genomic RNAs that were transcribed in vitro (data not shown).

View larger version (87K): [in this window] [in a new window]   FIG. 6.   Analysis of genomic RNA species extracted from infected liver under different experimental conditions. Identical samples of frozen liver tissue were allowed to thaw for 0, 1, 3, 10, and 30 min (lanes 1 to 5, respectively) prior to extraction. RNA samples were then subjected to Northern analyses (A) or primer extension to detect the 90-nt species indicative of opening at nt 1212 (B). The species were subjected to electrophoretic separations into 3% agarose (A) or 10% polyacrylamide-8 M urea (B), with size markers as indicated in lanes M.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have detected and characterized a novel opening site on the genome of HDV (Fig. 1). The 5' end so produced was determined by primer extension to be largely at nt 1210, with a weaker signal at nt 1211 (Fig. 2). In contrast, when we used an RLM-RACE procedure, we determined the end to be at nt 1212 (Fig. 3). We are confident that the discrepancy is due to a known weakness of primer extension, namely, that when the reverse transcriptase reaches the end of the template, one or more nontemplated nucleotides can be added (18). Consistent with this interpretation, we previously used primer extension to determine the 5' end of the HDV mRNA as nt 1631 ± 1 (7), and yet when we used a RACE procedure we deduced the site to be nt 1630 (5).

Using the RLM-RACE procedure, we also obtained evidence that the nature of the 5' end at nt 1212 was a 5'-OH (Fig. 3). Since such an end can also be produced by the HDV ribozyme (20), we carried out further experiments which we consider eliminate the possibility that the ribozyme was involved (Fig. 5). Other data support the interpretation that the opening happened in vivo and yet under conditions where only 4% of the unit-length antigenomic RNA was linear (Fig. 6). We have no explanation for why the opening was specific for genomic relative to antigenomic RNA.

This opening on genomic RNA at nt 1212 was observed for the liver RNA from each of the two infected animals studied. In other studies it was detected, but to a lesser extent, in transfected cultured cells; the amounts were highest at longer times (18 days) after the initiation of genome replication, but still these amounts were lower than those detected in the infected liver (data not shown). However, detection of this opening was somehow restricted to cells in which HDV was replicating. The genomic RNA isolated from the serum particles of an infected animal was predominantly circular in conformation (2), and the specific opening was not detectable on the linear forms (data not shown).

What then is the origin of this specific opening? One possibility is that it is a site of initiation of transcription. Navarro and Flores have recently described certain discrete 5' ends on unit-length viroid RNAs that have a triphosphate end and so may be sites of initiation (16). Also, we have recently reported that the 5' end of the HDV mRNA located at nt 1630 can have a cap structure (6), and so this also might be a site for the initiation of transcription. However, for the site we have described here, at nt 1212 on genomic RNA, there is neither a cap structure nor a triphosphate.

We consider it most likely that the 5' end at nt 1212 represents a specific endonucleolytic cleavage. Certainly many RNases cleave RNA by an intramolecular phosphoester transfer reaction to produce a fragment with a 5'-OH terminus (22). Moreover, there are numerous precedents for the degradation of a host RNA to begin with a cleavage at a preferred site (4, 13, 23). What may be more germane, studies with plant viroids have detected what are equivalent to specific opening sites on unit-length linear viroid RNAs (16, 19). From model studies it is clear that the primary sequence and structure of the RNA can contribute to observed site specificity (22). Thus, the predicted rod-like structure of the HDV genomic RNA in the vicinity, as shown in Fig. 7, along with the binding or lack of binding of host proteins, or of the HDV-encoded delta antigen, could contribute to produce the specificity that we have observed. In previous studies with transfected cells, we have shown that even in the absence of HDV genome replication, the presence of the delta antigen greatly increased the accumulation of processed HDV RNA circles (10, 15). It may also be relevant that the nucleotide sequences on the genomic RNA, both 5' and 3' of nt 1212, show that there are runs of pyrimidines (Fig. 7). However, as noted by others, there are many oligopyrimidine and oligopurine tracts on HDV RNAs (1).

View larger version (14K): [in this window] [in a new window]   FIG. 7.   Predicted folding of the rod-like genome at and around nt 1212. Note the oligopyrimidine stretches both 5' and 3' of the target site, which is indicated by the arrow. The sequence data and the predicted folding are from Kuo et al. (9).

If the opening was created by endonucleolytic cleavage, then the most obvious interpretation is that the precursor to the cleaved RNA is a circle. It needs to be pointed out that we think nt 1212 may not be the only opening site. While it seems to be a preferred and specific site, we expect that other sites are created in a less specific and/or less efficient manner. For example, we have seen other primer extension products on genomic sequences present in the infected liver (Fig. 4C and 6B). Since these sites were less efficient but more numerous in location, we favor that they could be explained by endonuclease activity.