INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Hepatitis delta virus (HDV) is a subviral human pathogen that increases disease severity in those infected with its helper, hepatitis B virus (29). The viral particle is composed of the single-stranded circular 1.7-kb HDV RNA, two forms of the viral protein, hepatitis delta antigen (HDAg), and hepatitis B surface antigen (2). HDAg binds specifically to HDV RNA (12, 20), and the short form, HDAg-S, is required for viral RNA replication (15). The mechanism of RNA replication is not known in detail, although several essential functional elements of HDAg, including the regions involved in HDAg dimerization (amino acids 12 to 60 [[30, 36]) and RNA binding (amino acids 89 to 145 [18, 20]), have been identified.

Analysis of HDV isolates from around the world has indicated that there are at least three phylogenetically distinct genotypes with different geographic distributions and associated disease patterns. Genotype I is the most widespread geographically (26, 31), having been identified in isolates from North America, Europe, Africa, east and west Asia, and the South Pacific, and is associated with a broad spectrum of chronic disease (26). Genotype II has been found only in east Asia and may be responsible for some of the milder forms of HDV disease in that region (33, 34). Genotype III is exclusively found in northern South America, where HDV infection is associated with particularly severe disease (7, 10, 25). Sequence divergence among the genotypes is as high as 40% over the entire RNA genome and 35% for the amino acid sequence of HDAg (7).

Thus far, the characterization of HDV genotypes has been based on phylogenetic analysis of sequences, and studies of HDV replication have been limited to clones of HDV genotype I isolates, most of which are closely related. In this report, we investigate replication of HDV genotype I and III clones in cultured cells to determine whether there might be functional as well as genetic differences between HDV genotypes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Plasmids for HDV genome expression. The genotype III expression plasmids pHDV·III(+) and pHDV·III() were created from cloned PCR amplification products from the isolate Peru-1 (7). Briefly, amplification products spanning nucleotides (nt) 307 to 1290 and 1078 to 424 of the circular 1,677-nt genome were cloned into the plasmid vector pT7Blue (Novagen) to yield clones pNV1-2 and pNV3-1, respectively. The 1-kb PstI fragment from pNV1-1 was cloned into the PstI site of pNV3-1 to yield the monomeric clone pNV3/1. The 1.7-kb KpnI fragment from pNV3/1 was cloned between the KpnI sites of clone pNV1-3, a sister clone to pNV1-2, to yield pPeru-1×1.5. From clone pNV1-3, the 466-bp SmaI fragment encompassing nt 522 to 988 of the Peru-1 genome was cloned into the SmaI site of pGEM-3Zf(+) (Promega) to yield clone p293. The 1,677-bp Esp3 I fragment from pPeru-1×1.5 (the Esp3 I site is at position 795 in the Peru-1 isolate) was cloned into the Esp3 I site of p293 to yield pPeru-1×1.2, which contained a duplication of both the genomic and antigenomic ribozyme autocleavage domains and encoded the short form of HDAg. The EcoRI-XbaI fragment from pPeru-1×1.2 was then inserted into the same sites of the plasmids pCMV2 and pCMV3 (8) to yield the expression plasmids pHDV·III(+) and pHDV·III(), which direct the synthesis of antigenomic and genomic HDV RNA, respectively, in transfected cells.

Plasmids for HDAg expression. The genotype I HDAg expression plasmid pCMV-AgS, which encodes the short form of HDAg, is described elsewhere (28). For clarity, we refer to this construct here as pHDAg-S·I. The expression construct pHDAg-L·I, which encodes the long form of genotype I HDAg, was created by substituting the PstI-SalI fragment of the plasmid pGDC-1M2 (4), which contains the nucleotide C at position 1012, for the same fragment in pHDAg-S·I. The genotype III expression plasmid pHDAg-S·III was obtained by substituting the 872-bp SstI-ApaI fragment from pPeru-1×1.2 for the same fragment in pHDAg-S·I. Plasmid pHDAg-S·III contains positions 779 to 1651 from the Peru-1 isolate and includes the coding sequence for HDAg and the polyadenylation signal. Expression plasmid pHDAg-L·III, which encodes the long form of genotype III HDAg, was obtained by substituting the 310-bp PstI-ApaI fragment from pNV1-3, which contains the nucleotide C at position 1014, for the same fragment in pHDAg-S·III.

Chimeric HDAg expression plasmids. The chimeric constructs pHDAg-S·III·III·I and pHDAg-S·I·I·III were obtained by switching 0.7-kb NgoA IV fragments between pHDAg-S·I and pHDAg-S·III. The NgoA IV site, located at position 1161 in the genotype I isolate, is identical in both isolates. Clones were screened by restriction digestion for the correct orientation. Plasmid pHDAg·III·I·I contained genotype III sequences from the beginning of the HDAg sequence up to the ApaI site in genotype I (nt 1380); it was created by a two-step procedure. Genotype III sequences between nt 1380 and 215 were amplified with primers 5'-TTTCCTGCCTCGGGCCCTCTTCGCC-3' and 5'-TTTGCTGCCGATGGGCCCTCGGACCGGGGTC-3' and cloned between the ApaI sites of the genotype I plasmid pGDC-1 to yield pG{Peru/DC-1} (Apa); clones were checked by restriction digestion for proper orientation. The primers used were specific for genotype III sequences, but internal ApaI sites were added; the ApaI site in the primer for the 1380 region is located such that the coding sequence is uninterrupted through this site when the genotype I and genotype III sequences are combined. The 601-bp BbsI-Esp3 I fragment from pG{Peru/DC-1}(Apa) was inserted between the same sites in plasmid pHDAg-S·III to yield pHDAg-S·III·I·I. Plasmid pHDAg-S·III·I·III, containing genotype I HDAg-S sequences between the ApaI and NgoA IV sites and genotype III sequences elsewhere, was created by substituting the 708-bp NgoA IV segment from pHDAg-S·III for the same fragment in pHDAg-S·III·I·I. Plasmid pHDAg-S·I·III·I, containing genotype III HDAg-S sequences between the ApaI and NgoA IV sites and genotype I sequences elsewhere, was created by a three-step procedure. The region between nt 1380 and 883 was amplified with primers 5'-GCGAAGAGGGCCCGGCAGGAAACCATG-3' and 5414 (26); the primer around 1380 contains an internal ApaI site which is located such that the coding sequence is uninterrupted through this site when genotype I and genotype III sequences are combined. The ApaI-PstI fragment of the PCR amplification product was substituted for the same fragment in plasmid pHDAg-S·III·I·I to give pHDAgP/C(Apa-Pst). The NgoA IV fragment of pHDAgP/C(Apa-Pst) was replaced by the same fragment of pHDAg-S·I to give pHDAgP/C(Apa-Ngo); the ApaI fragment of this plasmid was substituted for the same fragment in pHDAg-S·I to yield pHDAg-S·I·III·I. Plasmid pHDAg-S·I·III·III was created by substituting the NgoA IV fragment of pHDAg-S·III for that in pHDAg-S·I·III·I.

Transfections. Huh-7 cells were plated in six-well (35-mm-diameter) dishes and transfected with 5 µg of DNA per well by the calcium phosphate method as described previously (6). Transfections included the plasmid pSEAP2Control (Clontech), which was used to correct for transfection variability by monitoring secreted alkaline phosphatase expression (14). Transfection efficiencies varied by 2.5-fold or less. All experiments included duplicate transfections, and all were repeated at least twice.

RNA analysis. Cellular RNA was harvested as described previously (6) between days 5 and 7 posttransfection, as indicated, except that proteinase K (1 mg/ml) was included in the cell lysis buffer. For gel electrophoresis and blot hybridization analysis, 15% of the total RNA isolated was treated with DNase (Life Technologies) for 30 min prior to electrophoresis in 6.7% formaldehyde-1.5% agarose gels. Gels were stained with ethidium bromide and examined by UV illumination to verify equal loading and sample integrity. Samples were electroblotted onto Nytran (Boehringer Mannheim) membranes and fixed by UV illumination (Stratalinker; Stratagene, La Jolla, Calif.). Complete and even transfer of RNA samples to the membrane was verified by inspection of ethidium bromide-stained RNAs under UV illumination. Hybridization was for 16 to 36 h at 60°C in buffer containing 50% formamide, 1 M NaCl, 10% dextran sulfate, 1% sodium dodecyl sulfate (SDS), 0.5 mg of sheared salmon sperm DNA per ml, and hybridization probe (10⁶ cpm/ml; 5 × 10⁸ cpm/µg). Blots were washed with 0.1× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) at 65°C and then exposed to film or quantified by radioanalytic imaging (InstantImager; Packard Instruments, Meriden, Conn.). Hybridization probes were monomeric genotype I or genotype III RNAs, as indicated, except where noted otherwise. Genotype I probes were used to detect genotype I RNAs, and genotype III probes were used to detect genotype III RNAs. Under the stringent hybridization and washing conditions used, homologous hybridization is favored over heterologous hybridization because the sequences of genotypes I and III are only 60% identical overall, and the longest contiguous segment of sequence identity is 35 nt. However, it is possible to detect genotype III RNA with a genotype I RNA probe, and vice versa. For purposes of quantitation, hybridization signal intensities were corrected for transfection efficiency by cotransfection of a reporter expression construct for secreted alkaline phosphatase (14).

Immunoblot analysis. Cell lysates were obtained by treatment with 2% SDS-0.2 M Tris-HCl (pH 7.5)-1 mM EDTA and analyzed for hepatitis delta antigen by SDS-polyacrylamide gel electrophoresis in 12% acrylamide gels and immunoblotting with human monoclonal anti-HD T1/39 as described elsewhere (27).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

A replicating HDV genotype III cDNA clone. Analyses of HDV replication in cultured cells have thus far been limited to cDNA clones of HDV genotype I isolates. Two of the clones used (6, 15, 22) were derived from animals infected in related passage series and differed by about 1% (16, 32), a third (11, 23) was obtained from a genotype I isolate (24) that is about 10% different in nucleic acid sequence over the entire genome. Sequence variations among the genotypes, on the other hand, are as high as 40% (7), and inspection of sequence differences within some functional elements (such as the RNA editing site and the C terminus of the long form of HDAg) suggests that there could be important functional differences among the genotypes (3, 9). To compare the replication of genotypes I and III, we obtained a full-length cDNA clone from the prototype genotype III isolate Peru-1 (7). Expression constructs in which the cytomegalovirus promoter directs the synthesis of HDV RNA containing duplications of the ribozyme elements and encoding the short form of HDAg were created (Fig. 1). These constructs are similar to those used for analysis of HDV genotype I replication (6).

View larger version (18K): [in this window] [in a new window]   FIG. 1.   Schematic diagram of plasmid constructs used. Thick bars indicate HDV sequences; thin lines represent plasmid sequences. The open segments within the thick bars indicate sequences encoding HDAg. Dashed lines indicate HDV sequences which have been deleted. Numbers indicate sequence coordinates according Wang et al. (32) for genotype I and Casey et al. (7) for genotype III. Rightward-pointing arrows represent cytomegalovirus (CMV) immediate-early promoter. Downward-pointing filled arrowheads and upward-pointing open arrowheads indicate the positions of the antigenomic and genomic RNA autocatalytic cleavage sites, respectively. Ag() refers to HDAg-deficient RNAs; otherwise, () refers to genomic HDV RNA and (+) refers to antigenomic HDV RNA. Sequences shown directly beneath diagrams of pHDV·I(+)Ag() and pHDV·III()Ag() are for the first six and seven codons, respectively, of the genotype I and III HDAg-coding regions and indicate the position where a mutation was made by inserting a T. Plasmids pHDAg-L·I and pHDAg-L·III, described in Materials and Methods, are the same as their respective counterparts pHDAg-S·I and pHDAg-S·III except that they contain a single nucleotide change which alters the stop codon of HDAg-S from UAG to UGG (Trp) (21). Plasmid pHDV·I(+)Apa was described previously as pCMV2-DC-1×1(A) (8). (A) Expression constructs for wild-type genotype III HDV; (B) expression constructs for genotype I and III RNA defective for HDAg; (C) expression constructs for nonreplicating genotype I and III antigenomic RNAs. (D) expression constructs for HDAg.

View larger version (64K): [in this window] [in a new window]   FIG. 2.   RNA replication and HDAg production following transfection with clones of genotype I and genotype III HDV. Huh-7 cells were transfected with pCMV3-DC1×1.2 (left) or pHDV·III(+) (right), which express 1.2 unit-length genotype I and genotype III antigenomic RNA, respectively. Cells were harvested and RNA and proteins were isolated on days 6 (lanes 1 and 4), 9 (lanes 2 and 5), and 12 (lanes 3 and 6). (A) Northern blot analysis of genomic HDV RNA; (B) immunoblot analysis; (C) second run of samples from lanes 3 and 6 of panel B in adjacent lanes in the same gel.

Genotype-specific support of RNA replication by HDAg. In cultured cells, HDV RNA replication requires expression of the short form of HDAg (16). Thus, transfection of HDV cDNA constructs deficient in HDAg expression yields replicating HDV RNA only if an expression construct for HDAg is also cotransfected and if the defect which abrogates HDAg expression leaves intact structural elements required for RNA replication (1, 15, 17, 35). To examine the ability of HDAg to support replication of genotype I and genotype III HDV RNAs, constructs for intracellular expression of RNAs defective for HDAg synthesis were created (Fig. 1B). The defect was created by inserting a thymidine nucleotide in the cDNA at the beginning of the seventh and sixth codons, respectively, of the wild-type genotype I and genotype III expression constructs. This insertion created a stop codon as well as a frameshift. A single-base insertion was used in order to minimize potential effects on the RNA secondary structure that could also affect the ability of the RNA to replicate (35).

View larger version (26K): [in this window] [in a new window]   FIG. 3.   Genotype-specific support of RNA replication by HDAg-S of genotypes I and III. Huh-7 cells were transfected with indicated constructs for the expression of genomic (left) or antigenomic (right) HDV RNA defective for HDAg production. Cells were also cotransfected with the plasmid vector pCMV3 (lanes ) or plasmids pHDAg-S·I (lanes I) and pHDAg-S·III (lanes III), which express genotype I and III HDAg, respectively. RNAs were isolated 6 days posttransfection and analyzed by Northern blotting. Hybridization probes detected antigenomic RNA (left) or genomic RNA (right).

HDAg clones from different HDV isolates. The results presented in Fig. 2 and 3 were obtained with clones derived from the prototype isolates of HDV genotypes I and III. To determine whether the observed effects might be due to genotype-specific differences rather than a peculiarity of the two isolates used, additional HDAg expression clones were obtained by cloning PCR-amplified cDNAs. Three clones were obtained from additional unrelated genotype I isolates and four from additional unrelated genotype III isolates. Within a 357-bp region (nt 908 to 1265) corresponding to the C-terminal half of the HDAg-coding region, extents of sequence divergence among these genotype I and genotype III isolates were from 6 to 9% and from 4 to 8%, respectively (4, 10, 26, 31). Upon cotransfection of Huh-7 cells, all four genotype I HDAg clones supported replication of HDAg() genotype I RNA, and all five genotype III HDAg clones supported replication of HDAg() genotype III RNA (Fig. 4). However, none of the HDAg clones was able to support replication of HDAg() RNA of the other genotype at more than 5% of the level supported by the prototype HDAg clones of the same genotype (Fig. 4). These results were verified in three separate transfection experiments. Variations in transfection efficiency, as determined by analysis of a cotransfected reporter gene (14), were less than twofold and could not account for the different levels of RNA replication.

View larger version (50K): [in this window] [in a new window]   FIG. 4.   Genotype-specific support of RNA replication is exhibited by HDAg clones from multiple genotype I and genotype III isolates. Huh-7 cells were transfected with plasmids pHDV·I(+)Ag() (top) and pHDV·III(+)Ag() (bottom), which express antigenomic HDV RNA defective for HDAg production. Cells were also cotransfected with expression constructs for HDAg clones from different isolates (lanes 1 to 4, genotype I; lanes 5 to 9, genotype III) or with the plasmid vector pCMV3 (lanes 10). Cotransfected clones: lanes 1, genotype I prototype clone pHDAg-S·I; lanes 2, isolate U02 (7); lanes 3, isolate I27 (26); lanes 4, isolate I43; lanes 5, genotype III prototype clone pHDAg-S·III; lanes 6, isolate Peru 5 (10); lanes 7, isolate Peru 13 (10); lanes 8, isolate P20; lanes 9, isolate P21. RNAs were isolated 7 days posttransfection and analyzed by Northern blotting for HDV genomic RNA. Numbers beneath lanes indicate RNA levels quantified by radioanalytic imaging and normalized for variations in transfection efficiency by evaluating expression of a cotransfected reporter gene (14). Values shown are relative to the prototype (lane 1 in the upper panel; lane 5 in the lower panel), which is assigned a value of 100, and are the averages of two transfections.

Stabilization of HDV RNA by HDAg. HDAg has been shown to bind specifically to HDV RNA, and this binding is required for HDV RNA replication (19). In transfected cells, the interaction between HDAg and HDV RNA has also been shown to stabilize HDV RNAs that contain large deletions which preserve the ability of the RNA to form the characteristic unbranched rod structure but abolish the replication activity of the RNA (17). To assess whether inefficient interactions between RNAs and HDAg of different genotypes could be responsible for the diminished ability to support replication, Huh-7 cells were cotransfected with pHDAg-S·I or pHDAg-S·III and either pHDV·I(+)Apa or pHDV·III(+)BX. The latter two constructs direct the synthesis of antigenomic RNAs containing internal deletions that preserve the ability of the RNA to form the predicted unbranched rod structure but prevent RNA replication even in the presence of HDAg. Six days posttransfection, RNAs were harvested and analyzed by Northern blotting for levels of antigenomic transcripts transcribed from pHDV·I(+)Apa or pHDV·III(+)BX. The hybridization probes did not contain HDAg coding sequences and did not detect appreciable RNA in cells transfected with either pHDAg-S·I or pHDAg-S·III alone (Fig. 5).

View larger version (34K): [in this window] [in a new window]   FIG. 5.   Stabilization of HDV RNA by HDAg. Huh-7 cells were transfected with the indicated expression constructs for HDV antigenomic RNA. Cells were also cotransfected with the plasmid vector pCMV3 (lanes ) or plasmids pHDAg-S·I (lanes I) and pHDAg-S·III (lanes III), which express genotype I and III HDAg, respectively. RNAs were isolated 6 days posttransfection and analyzed for HDV antigenomic RNA by Northern blotting; the hybridization probes did not include HDAg-encoding sequences transcribed by the HDAg expression constructs.

Chimeric HDAg constructs. Comparison of the predicted amino acid sequences of genotype I and III HDAg indicates both conserved and variable regions (Fig. 6A). Studies of genotype I HDAg have identified several functional elements, including a dimerization domain (30, 36), nuclear localization signals (36), and a bipartite RNA-binding domain containing two arginine-rich motifs (19). To attempt to identify regions and possible functional elements responsible for the genotype-specific support of HDV RNA replication, chimeric HDAg constructs containing mixtures of type I and type III sequences were created and analyzed for the ability to support replication of type I and type III RNAs. The HDAg coding region was divided into three approximately equal-size regions, which were exchanged between the genotype I and III clones to create a series of six chimeric HDAg expression constructs (Fig. 6B). The amino-terminal third included the dimerization domain, the middle third included the RNA-binding region, and the C-terminal third included a glycine-proline-rich region of unknown function. Amino acid sequence identities were 60, 70, and 78% for the N-terminal, middle, and C-terminal segments, respectively. Huh-7 cells were cotransfected with these chimeric HDAg expression constructs and the HDAg() RNA expression construct pHDV·I(+)Ag() or pHDV·III(+)Ag(), each of which directs expression of HDAg() antigenomic RNA. Seven days posttransfection, RNA was harvested and analyzed for the presence of HDV genomic RNA by Northern blot hybridization. RNA levels were quantified by radioanalytic imaging and are presented relative to that observed for cotransfection with the prototype HDAg constructs (Fig. 6).

View larger version (33K): [in this window] [in a new window]   FIG. 6.   Support of RNA replication by chimeric genotype I-genotype III HDAg expression constructs. (A) Comparison of HDAg-S sequences of genotype I and III prototype isolates. Uppercase letters for consensus indicate agreement between genotype I and III clones. Dashes in the consensus indicate positions not in agreement; for each isolate, the identities of these positions are indicated in lowercase. Dots indicate gaps introduced to produce an optimal alignment. Lines above sequences indicate boundaries used in the creation of the chimeric constructs (B). Functional domains identified in other studies are indicated in open boxes beneath sequences. The black boxes within the RNA-binding domain indicate arginine-rich motifs (19). (B) Left, schematic representation of HDAg chimeric constructs. Open boxes indicate genotype I sequences, shaded boxes indicate genotype III sequences. Right, relative levels of RNA replication supported by chimeric HDAg constructs. Huh-7 cells were transfected with pHDV·I(+)Ag() (genotype I RNA) or pHDV·III(+)Ag() (genotype III RNA) and with the indicated HDAg expression constructs. RNAs were isolated 7 days posttransfection and analyzed by Northern blotting for the appropriate genomic HDV RNA. RNA levels were quantified by radioanalytic imaging and normalized for expression of a cotransfected reporter gene (14). Values indicated are relative to the levels of expression obtained with the homologous prototype HDAg clone, as described for Fig. 4. Quantitations presented are averages from two independent transfections and did not vary by more than 10% in separate experiments. Relative activities that were less than 1% of the prototypes are shown as 0.

Effects of HDAg on replication of wild-type HDV. The inability of HDAg from genotypes I and III to fully support replication of HDAg() RNA from the other genotype raised the question of what effects might occur on replication of wild-type HDV. Huh-7 cells were cotransfected with HDV expression constructs for genotype I and genotype III and expression constructs for the short form of HDAg from both genotypes; for comparison, expression constructs for the long form of HDAg were also cotransfected independently. For wild-type genotype I, cotransfection of an expression construct for HDAg-S(I) had little effect on RNA replication, and as expected (15), coexpression of the long form of HDAg suppressed replication (Fig. 7). Surprisingly, coexpression of the short form of genotype III HDAg strongly inhibited genotype I RNA replication; remarkably, this inhibition was even greater than that due to the long form of genotype I HDAg (Fig. 7). This inhibition was not limited to the prototype genotype I isolate; RNA replication of another genotype I isolate (clone L [5]; GenBank accession no. L22066) was also strongly inhibited by coexpression of genotype III HDAg (4). The strongest suppression of genotype I RNA replication was observed with the long form of genotype III HDAg. Genotype III RNA replication was less affected by the presence of genotype I HDAg. Cotransfection of pHDAg-S·I with pHDV·III(+) resulted in a two- to threefold inhibition of replication, while pHDAg-S·III increased replication of the wild-type genotype III RNA. However, genotype I HDAg-L strongly inhibited replication of genotype III RNA, as did genotype III HDAg-L.

View larger version (43K): [in this window] [in a new window]   FIG. 7.   Inhibition of genotype I RNA replication by genotype III HDAg. Huh-7 cells were transfected with pCMV3-DC1×1.2 (left) or pHDV·III(+) (right), which express wild-type 1.2 unit-length genotype I and genotype III antigenomic RNA, respectively. Cells were also cotransfected with the indicated HDAg expression constructs or with the plasmid vector pCMV3 (lanes None). RNAs were isolated 6 days posttransfection and analyzed by Northern blotting for HDV antigenomic RNA.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Previous studies of HDV replication have been limited to three different clones of HDV genotype I, two of which are nearly identical. In this report, we find that the roles of the short and long forms of HDAg in viral RNA replication are similar for HDV genotypes I and III, despite sequence variations of about 40%. The short form of HDAg is required for RNA replication (Fig. 3), while the long form inhibits RNA replication (Fig. 7). Moreover, for both genotypes, HDAg can bind HDV RNA of the same type in cells, as evidenced by the stabilization of the RNA in the presence of HDAg (Fig. 5). It is notable that the inhibitory activities of HDAg-L are similar for the two types (Fig. 7), even though the sequence of the C-terminal extension responsible for this inhibitory activity (13) is very different in genotypes I and III (7, 10, 31).

Despite these similarities, however, we describe important functional distinctions between HDV genotypes I and III that further confirm the original classification of sequences within these phylogenetically distinct clades as belonging to unique genotypes (7, 26, 31). The observation that HDAg clones from genotype I and III isolates cannot fully support replication of HDV RNA from the other genotype (Fig. 2, 4, and 6) strongly suggests that the support of replication by HDAg is genotype specific. Although the wide variation of genotype I RNA replication supported by genotype I HDAg clones raises the formal possibility that there are also isolate-specific effects (within a genotype), it seems unlikely that the genotype-specific effects observed are simply an extension of an isolate-specific effect which is related to genetic similarities. There was no correlation between sequence divergence and the amount of replication supported by the different genotype I HDAg clones. The relative levels of genotype I RNA replication supported by HDAg clones from isolates U02, I27, and I43 were 1.0, 0.2, and 11, respectively (Fig. 4), while the sequence similarities between these isolates and the prototype were 6.5, 7.3, and 9.6%, respectively (4, 26, 31). Moreover, notwithstanding possible isolate-specific effects, the genotype-specific effect was clear: all HDAg clones supported replication of the prototype RNA from the same genotype substantially better than the prototype RNA of the other genotype, regardless of their activities on the RNA of the same genotype (Fig. 4).

The genotype-specific activity of HDAg-S from genotypes I and III in supporting HDV RNA replication (Fig. 3, 4, and 6) suggests that these clades may best be referred to as types rather than genotypes (i.e., HDV-I and HDV-III), the distinction being that the description of the uniqueness of the different sequence groups is not limited to genetic analysis but includes functional activity essential to the virus life cycle. In this regard, it is notable that the capsid and polymerase proteins of hepatitis B virus and woodchuck hepatitis virus, which infect different hosts, have been shown to be functionally interchangeable (37). The different electrophoretic mobilities of genotype I and III HDAg, which could be due to differences in posttranslational modifications (e.g., phosphorylation) or structural variations, further confirm the differences between these genotypes and may be related to the different functional activities of the two species. Because the regions of sequence variation between HDAg of HDV-I and HDV-III also vary in genotype II (7), it seems likely that the genotype-specific support of replication will also apply to HDV genotype II. Whether such is the case can readily be determined after full-length clones of HDV genotype II are available.

The mechanistic basis for the inability of HDAg-S of either HDV-I or HDV-III to support replication of RNA of the other type is uncertain, but some possibilities are suggested by the data. Others have shown that the ability of HDAg-S to support HDV RNA replication requires the RNA-binding domain of HDAg (18, 19). Type III HDAg was apparently less effective at stabilizing type I RNA than was type I HDAg (Fig. 5), suggesting a weaker binding interaction between the heterologous pair than between the homologous RNA and HDAg species. Thus, the inability of type III HDAg to support type I RNA replication could be due, in part, to a weak interaction between this HDAg-RNA pair. However, this explanation does not appear to apply to the inability of type I HDAg to support type III RNA replication, because type I HDAg apparently binds to and stabilizes type III RNA (Fig. 5). Perhaps the specific mechanisms by which the short form of HDAg facilitates RNA replication are not the same for genotypes I and III.

The nonreciprocal nature of activities of type I and III HDAg in stabilizing type I and type III RNAs is curious. This stabilization is most likely due to binding between HDAg and HDV RNA, as previously reported (17). Both type I and type III RNAs used in the experiments presented in Fig. 5 are predicted to form unbranched rod structures with extensive base pairing, and most of the sequences within the arginine-rich motifs are highly conserved between types I and III (Fig. 6). At least for genotype I, these features have been shown to be required for binding between HDAg and HDV RNA (12, 19). However, the specific determinants of binding have not been identified in either the RNA or HDAg. Further comparative analysis of the binding activities of the RNAs and HDAgs of types I and III is likely to prove instructive for understanding these determinants (on both HDAg and HDV RNA) and their roles in HDV RNA replication.

The broad range (about 20-fold) of replication supported by HDAg from different genotype I isolates was remarkably greater than that observed among genotype III isolates, particularly as the extents of sequence variation in a segment including the C-terminal half of the antigen coding region were similar among the two groups of isolates. As discussed above, it is possible that variations in the level of HDV genotype I prototype RNA replication supported by the different genotype I HDAg clones were due to isolate-specific interactions between HDAg and HDV RNA. Alternatively, the wide range of replication supported by the different type I isolates could be due to differences in the inherent abilities of HDAg from these isolates to support replication. Indeed, viremia varies considerably among different patients (26), and it is tempting to speculate that the broad spectrum of disease associated with type I HDV infection (26) is due, in part, to variations in the ability of HDAg in different isolates to support high levels of RNA replication. Analyses of the abilities of HDAg clones to support replication of HDV RNAs from additional isolates, particularly those with divergent sequences and/or disease and viremia patterns, will be useful in examining these possibilities.

The strong inhibition of genotype I RNA replication by genotype III HDAg could have implications for our understanding of both the molecular biology and the epidemiology of this agent, as well as for the development of novel therapeutic approaches. Because genotype III HDAg-S bound genotype I RNA only weakly (Fig. 5), it seems unlikely that the mechanism of inhibition would involve sequestration of the genotype I RNA by genotype III HDAg. An alternative possibility involves direct interaction between HDAg-S of genotypes I and III such that the replication activity of genotype I HDAg is inhibited. In this event, the lack of inhibition of genotype III RNA replication by genotype I HDAg could be explained if the genotype I-III HDAg complex alters the conformation or accessibility of the N-terminal region of genotype I HDAg, which is required for genotype I RNA replication but has little effect for genotype III (Fig. 6). Regardless of the mechanism, the inhibition of HDV-I by type III HDAg raises the interesting possibility that in the natural setting, genotype III may have a competitive advantage in a type I-type III coinfection. Conceivably, further analysis of this inhibitory activity could be used to develop therapies to inhibit genotype I HDV replication in infected patients.