307 Current Topics in Microbiology and Immunology

Transcription

1 307 Current Topics in Microbiology and Immunology Editors R.W. Compans, Atlanta/Georgia M.D. Cooper, Birmingham/Alabama T. Honjo, Kyoto H. Koprowski, Philadelphia/Pennsylvania F. Melchers, Basel M.B.A. Oldstone, La Jolla/California S. Olsnes, Oslo P.K. Vogt, La Jolla/California H. Wagner, Munich

2 J.L. Casey (Ed.) Hepatitis Delta Virus With 25 Figures and 12 Tables 123

3 John L. Casey, Ph.D. Department of Microbiology and Immunology Georgetown University 3900 Reservoir Road, NW Washington, DC USA The cover illustration is a simplified structure of hepatitis delta virus showing the internal ribonucleoprotein complex, which contains the circular RNA genome and the two forms of the hepatitis delta antigen; the envelope proteins of hepatitis B virus form the exterior of the virus. The inset is an electron micrograph of purified hepatitis delta virus particles, and was kindly provided by Dr. John Gerin. The background immunofluorescence image is of transfected cells expressing hepatitis delta antigen, and was kindly provided by Dawn Defenbaugh. Library of Congress Catalog Number ISSN X ISBN Springer Berlin Heidelberg New York ISBN Springer Berlin Heidelberg New York This work is subject to copyright. All rights reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,reproductiononmicrofilmorinanyotherway,andstorageindatabanks.duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September, 9, 1965, in its current version, and permission for use must always be obtained from Springer-Verlag. Violations are liable for prosecution under the German Copyright Law. Springer is a part of Springer Science+Business Media springeronline.com Springer-Verlag Berlin Heidelberg 2006 Printed in Germany The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Product liability: The publisher cannot guarantee the accuracy of any information about dosage and application contained in this book. In every individual case the user must check such information by consulting the relevant literature. Editor: Simon Rallison, Heidelberg Desk editor: Anne Clauss, Heidelberg Production editor: Nadja Kroke, Leipzig Cover design: design & production GmbH, Heidelberg Typesetting: LE-TEXJelonek,Schmidt&VöcklerGbR,Leipzig Printed on acid-free paper SPIN /3150/YL

4 Preface Since its discovery nearly 30 years ago, hepatitis delta virus (HDV) has continued to surprise and fascinate. Initially thought to be an antigenic variant of hepatitis B virus (HBV), HDV was soon found to be a defective virus that depends on an underlying or simultaneous hepatitis B infection. The clinical significance of HDV infection is more severe acute and chronic liver disease than that caused by the HBV infection alone. The cloning and sequencing of the genome led to the realization that HDV is a unique RNA virus whose closest known relatives are plant viroids, but even that relationship is remote. In the current classification scheme of the International Congress on the Taxonomy of Viruses, HDV remains the sole member of a floating genus, Deltavirus. The genome and its replication cycle bear no discernable resemblance to its helper virus, HBV, on which HDV depends for its envelope. At 1,680 nucleotides the HDV genome is the smallest known to infect man. The virus contains just one gene, which encodes an approximately 25-kDa protein, hepatitis delta antigen (HDAg, also sometimes referred to as delta protein or delta antigen). To compensate for this limited protein coding capacity, HDV relies heavily on host functions and on the structural dynamics of its circular RNA genome. Although HDV RNA is circular, it forms a characteristic unbranched rod structure in which over 70% of the nucleotides from Watson Crick base pairs. One of the more remarkable aspects of HDV is that, unlike other RNA viruses, it does not produce a virally-encoded polymerase; rather, it somehow uses host DNA-dependent RNA polymerase to replicate its RNA genome and transcribe its mrna. At a minimum, this process involves RNA polymerase II; HDAg also plays an as yet undefined role. The potential involvement of another polymerase, such as polymerase I, or of other forms of RNA polymerase II remains an area of active investigation. RNA replication requires the unbranched rod structure of HDV RNA and occurs via a double rolling circle mechanism. Autocatalytic self-cleaving elements, termed ribozymes, in the genome and its complement, the antigenome, play essential roles in the processing of linear transcripts to circular forms. Ribozyme activity occurs via

5 VI Preface acid-base catalysis not unlike that accomplished by protein enzymes, and requires a complex double pseudoknot RNA structure. Ribozyme activity is also controlled by the structural dynamics of the RNA: formation of the unbranched rod structure interferes with ribozyme activity and likely prevents cleavage from occurring once the RNA circularized. HDV produces two forms of HDAg that have different roles in the replication cycle. The longer form has an additional 19 or 20 C-terminal amino acids that facilitate viral particle formation; the shorter form is required for RNA replication. The heterogeneity arises due to highly specific editing of an adenosine in the antigenome RNA by host RNA adenosine deaminase. This process requires particular secondary structure features in the RNA around theeditingsite.insomecasestheunbranchedrodstructurecompeteswiththe configuration required for editing; thus, structural dynamics of the RNA are important not only for HDV ribozyme activity, but for other processes as well. ThefunctionalactivityofHDAgisaffectedbynumerouspost-translational modifications carried out by host enzymes. These modifications include farnesylation, phosphorylation, methylation and acetylation. Farnesylation is essential for interaction with the hepatitis B virus surface protein (HBsAg), and is thus required for viral particle formation. The specific significance of the other modifications, as well as the nature of their effects on HDAg function, are not yet fully understood. Being derived from HBV proteins, the outside of the HDV particle is similar to that of HBV, only slightly smaller in size. Although the receptor for neither virus has been identified it is likely that attachment and entry occur by similar processes. Infectivity of both HBV and HDV involves elements of the pres1 and antigenic loop regions of HBsAg. Molecular genetic analysis of HDV isolates indicates geographical correlations that in some ways mirror those of its helper virus. That the greatest sequence diversity is found among isolates originating in Africa has led to the proposal that HDV might have radiated from that continent. One enigma is that the most distantly related sequences, for both HDV and HBV, come from South America. There is some evidence that infection with certain genotypes, or clades, can influence the severity of HDV disease. The mechanisms by which HDV thwarts the immune system to produce chronic infection are not yet understood. The woodchuck model of HDV has been the most accessible animal model of HDV infection and has been used both to analyze the natural history of HDV infection and to evaluate the efficacy of vaccine strategies against the virus. Certainly, development of an effective vaccine strategy has been frustrating. Recent work suggests that HDAg may be poorly immunogenic, and may furthermore undergo genetic changes to avoid those limited immune responses that do occur.

6 Preface VII There are no effective licensed antiviral therapies for HDV, and although several therapies exist for combating its helper, HBV, none of these treatments affect HDV. This failure is due to the fact that HDV depends only on HBsAg production of the helper, and current anti-hbv therapies are not potent enough to significantly diminish HBsAg levels, which are extraordinarily high. However, two potential therapeutic approaches have shown promise. One targets the host farnesyltransferase activity, which is required for virus production; the other approach advocates reducing HBsAg to levels that are too low to support continued HDV secretion. Both of these approaches are based to varying degrees on an understanding of the molecular virology of HDV, and it is likely that additional therapeutic avenues will be opened as our knowledge of HDV expands. The more we continue to learn about hepatitis delta virus the more fascinating it becomes. It is my hope that this book will stimulate additional interest in hepatitis delta virus among scientists, academic researchers and advanced students. I would like to thank the authors for their contributions, and the staff at Springer and members of my laboratory for their assistance in preparing this volume. Washington, DC, March 2006 John L. Casey

7 List of Contents StructureandReplicationofHepatitisDeltaVirusRNA... 1 J. M. Taylor HDVRNAReplication:AncientRelicorPrimer? T.B.MacnaughtonandM.M.C.Lai HDVRibozymes M. D. Been RNAEditinginHepatitisDeltaVirus J. L. Casey Post-translationalModificationofDeltaAntigenofHepatitisDVirus W.-H. Huang, C.-W. Chen, H.-L. Wu, and P.-J. Chen TheRoleoftheHBVEnvelopeProteinsintheHDVReplicationCycle C. Sureau Prenylation of HDAg and Antiviral Drug Development J. S. Glenn Hepatitis Delta Virus Genetic Variability: FromGenotypesI,II,IIItoEightMajorClades? P. Dény Functional and Clinical Significance ofhepatitisdvirusgenotypeiiinfection J.-C. Wu ImmunologyofHDVInfection M. Fiedler and M. Roggendorf TheWoodchuckModelofHDVInfection J. L. Casey and J. L. Gerin Subject Index...226

8 List of Contributors (Addresses stated at the beginning of respective chapters) Been, M. D. 47 Casey, J. L. 67, 211 Chen, C.-W. 91 Chen, P.-J. 91 Dény, P. 151 Fiedler, M. 187 Gerin, J. L. 211 Glenn, J. S. 133 Lai, M. M. C. 25 Macnaughton, T. B. 25 Roggendorf, M. 187 Sureau, C. 113 Taylor, J. M. 1 Wu, H.-L. 91 Wu, J.-C. 173 Huang, W.-H. 91

9 CTMI (2006) 307:1 23 c Springer-Verlag Berlin Heidelberg 2006 Structure and Replication of Hepatitis Delta Virus RNA J. M. Taylor ( ) Fox Chase Cancer Center, 333 Cottman Avenue, Philadelphia, PA , USA john.taylor@fccc.edu 1 Introduction RNAs and Ribonucleoproteins GenomeandAntigenome mrna OtherHDVRNAs RNAStructure Ribonucleoproteins RNA Replication RolesofDeltaProteins TheEssentialSmallDeltaProtein OtherFormsofDeltaProtein Replication with an Unchanging Separate Source of Small Delta Protein EnzymologyofRNA-DirectedTranscription InitiationofReplication DoubleRolling-CircleModel Template-Switching,Reconstitution,andRecombination InhibitionofReplication ResistancetoInterferons SensitivitytoRibavirin SensitivitytosiRNA ResistancetoDicer CytopathicEffectofReplication Evolution of the RNA Sequence AccumulationofChanges ADAR-Editing Origin Outlook References Abstract Whilethisvolumecoversmanydifferentaspectsofhepatitisdeltavirus (HDV) replication, the focus in this chapter is on studies of the structure and replication of the HDV RNA genome. An evaluation of such studies is not only an integral part of our understanding of HDV infections but it also sheds new light on some

10 2 J.M.Taylor important aspects of cell biology, such as the fidelity of RNA transcription by a host RNA polymerase and on various forms of post-transcriptional RNA processing. Representations of the replication of the RNA genome are frequently simplified to a form of rolling-circle model, analogous to what have been described for plant viroids. One theme of this review is that such models, even after some revision, deceptively simplify the complexity of HDV replication and can fail to make clear major questions yet to be solved. 1 Introduction Other reviews on the topic of hepatitis delta virus (HDV) RNA structure and replication have been previously published (Cunha et al. 2003; Gerin et al. 2001; Taylor 2003, 2004). Moreover this volume contains current reviews on other aspects of HDV infection, in addition to one chapter on HDV replication (see chapter by T.B. Macnaughton and M.M.C. Lai, this volume). The objective of this chapter therefore will be to not only review information regarding HDV genome structure and replication, but also to consider what might be new insights and to point to questions yet to be solved. Over the years HDV has provoked interest because of the many unique features associated with its replication including RNA-directed transcription by a host enzyme, ribozyme domains, and essential RNA editing. It has also been associated with a deceptive simplicity: a very small genome encoding only one or two viral proteins to account for, and a rolling-circle model of replication that seems plausible. However, as described in this chapter, we are becoming aware of a greater complexity associated with the replication of this apparently simple virus. 2 RNAs and Ribonucleoproteins 2.1 Genome and Antigenome The genome of HDV is a small RNA of about 1,700 nucleotides in length with a circular conformation. We often refer to this RNA as single-stranded; however, based on predictions from the nucleotide sequence and certain experimental studies, we are convinced that this RNA can fold on itself via intra-molecular base pairing to form an unbranched rod-like structure. In this way, about 74% of all the nucleotides are involved in base pairing (Sect. 2.4).

11 Structure and Replication of Hepatitis Delta Virus RNA 3 The HDV genome, by definition, is the RNA species that is incorporated into new virus particles during an assembly process that depends upon envelope proteins provided by the helper hepadnavirus (see chapter by C. Sureau, this volume). However, inside a cell undergoing HDV genome replication, in addition to the genomic RNA, there are also many copies of an exactly complementary RNA, referred to as the antigenome (Chen et al. 1986). This antigenomic RNA contains an open reading frame for a 195-amino acid species, the small delta protein (S-HDAg), which is essential for genome replication, is apparently not translated from the circular antigenomic RNA, but from a less than genome-sized, polyadenylated RNA species that is present in the cytoplasm (Sect. 2.2). Both the genome and antigenome contain a domain that will act as a ribozyme. As discussed more fully in the chapter by M.D. Been (this volume), these domains are as short as 85 nucleotides in length (Ferre-D Amare et al. 1998). They are sufficient to allow RNA cleavage in vitro, in the presence of magnesium ions. The cleavage is a site-specific trans-esterification reaction which produces a 5 -OH and a cyclic 2 -, 3 -monophosphate (Kuo et al. 1988b). This cleavage ability is needed for HDV replication (Macnaughton et al. 1993) and is considered to provide post-transcriptional cleavage of greater than unit-length HDV RNA multimers, thus releasing unit-length linear species that can be subsequently ligated to form new RNA circles (Taylor 1990). As considered in Sect. 2.4, the structures of the HDV ribozymes are different from the predicted rod-like folding. 2.2 mrna The mrna species for S-HDAg contains little more than the open reading frame. It is 5 -capped and 3 -polyadenylated, just as for a typical host mrna. Recent studies indicate that S-HDAg mrna can be bound with an antibody that can recognize 5 -cap structures (Nie et al. 2004). It is also bound by the poly(a) binding protein, PABP, a host protein that binds to the poly(a) sequences of host mrnas (X. Nie, J. Chang, C. Tarn, C.-M. Chiang, J. Keene, L. Penalva and J. Taylor, unpublished results). For the mrna there is evidence that the 5 -end has a preferred location at nucleotide 1630 (Gudima et al. 1999). At least for the majority of this mrna the 5 -end has been modified to have a cap structure (Gudima et al. 2000; Nie et al. 2004). Therefore it is plausible but not directly proven, that this 5 -end corresponds to a preferred site for the initiation of RNA-directed transcription from a genomic RNA templates (Sect. 3.3).

12 4 J.M.Taylor At the 3 -end of the mrna is a poly(a) tail of about 150 adenosines, typical of cellular mrna species. Consistent with this, the nascent RNA that is processed to become this mrna, has a poly(a) signal, AAUAAA, and other features expected for an RNA transcript that undergoes polyadenylation. Mutation of this signal inhibits polyadenylation (Nie et al. 2004). 2.3 Other HDV RNAs It should be clear from the above that all three of the major HDV RNAs arise by post-transcriptional RNA processing. This means that for each, there are precursor RNAs, of relatively larger size, that act as the substrates for such processing. There is much more yet to be revealed about how such important precursor RNA species are initiated and how they are processed. 1. As discussed in Sect. 3.4, an unsubstantiated aspect of the rolling-circle modelofreplicationiswhetherthernaspeciesthatgoontobeprocessed into antigenomic RNA circles are initiated from the same location as those which become mrna species. 2. For the nascent genomic RNA transcripts we have as yet no clear data as to where they are initiated. One report has suggested, based on RNase protection assays, that there might be a preferred site, near one end of the rod-like RNA, and almost opposite to the site that corresponds to the 5 - end of the mrna (Beard et al. 1996). This result needs to be independently confirmed. 3. Since the nascent transcripts of both the genomic and antigenomic RNA each contain a ribozyme domain per unit length, a transcript that is greater than unit length can contain two or more ribozyme domains, which will lead to ribozyme processing, to release RNA species that are of exactly unit length, and that can then be converted to unit length RNA circles. There is good evidence that the ribozyme domains are needed for the cleavage events to release the unit length linear RNAs (Macnaughton et al. 1993). There is also a report that the subsequent conversion of these to circles depends upon host factors, probably a host RNA ligase (Reid and Lazinski 2000). It will be important to identify this host ligase and determine how it is redirected to act on HDV RNAs. 4. During HDV replication additional processed RNAs that are relatively less abundant can be detected. Northern analyses can detect molecules that seem to be of twice or even three times unit length. Moreover, these species seem to exist in both circular and linear conformations (Chen et al. 1986). Apparently these species arise via alternative processing of multimeric nascent RNA transcripts.

13 Structure and Replication of Hepatitis Delta Virus RNA RNA Structure It was promptly deduced from the first full sequence of an HDV genome that this RNA could theoretically be folded into an unbranched rod-like structure (Wang et al. 1986). Further theoretical calculations predicted a high negative free energy of 805 kcal/mol, consistent with a stable structure (Kuo et al. 1988a). Experimental evidence also supports such a structure. In electron microscopic studies, the genomic RNA appears as a short double-stranded rod that upon progressive denaturation opens into a circle (Kos et al. 1986). In addition, by gel electrophoresis under nondenaturing conditions, both the genomic and antigenomic RNAs migrate as expected for double-stranded species (Lazinski and Taylor 1995). Moreover, upon prior denaturation, most of the RNAs migrate consistent with a circular conformation, the remainder behaving as linear species (Chen et al. 1986). Similar conclusions apply for the structure of the unit-length antigenome. While the rod-like folding is generally true, the details of the exact folding remain to be determined. One study attempted to use nuclease susceptibility assays to test the folding of a segment at one end of the rod-like structure of the genomic RNA in vitro. The detected folding was very close to that predicted (Beard et al. 1996). It is obvious that the structures of the two ribozyme domains are not compatible with the rod-like folding. Furthermore, folding of these domains into the rod-like structure should inhibit the ribozymes. Intuitively, if this inhibition were not the case the circular RNAs might undergo efficient self-cleavage to form linear RNAs. The prediction that the rod-like folding overrides the ability to fold into the active ribozyme conformation has been proven using both in vivo and in vitro studies of natural and modified HDV RNAs (Lazinski and Taylor 1993, 1994a). Modified RNAs that cannot fold the ribozyme domain into a rod-like structure do not form stable circles in vivo. Conversely, unmodified HDV RNA circles are cleaved only inefficiently by ribozymes in vitro. And yet if these RNAs are first hybridized with a separate oligonucleotide to stop the ribozyme domain from being inactivated by being drawn into the rod-like folding, then these RNAs are efficiently cleaved by ribozymes in vitro. Additional alternative foldings of HDV RNA sequences have also been reported. One such pairing is considered to produce a binding site for the protein PKR (Circle et al. 2003, 1997; Robertson et al. 1996). Another alternative folding, also based on in vitro data, has been proposed to explain a specific cross-linking induced by irradiation with ultraviolet light (Branch et al. 1989).

14 6 J.M.Taylor In summary, it should not be surprising that HDV RNAs can, and do have, multiple ways in which they can fold. It is considered that most RNAs undergo a series of alternative foldings, that is, metastable states that can facilitate the molecule ultimately achieving a predominant final more stable structure (Uhlenbeck 1995). On top of this, as will be considered in Sect. 2.5, the final structure for HDV genomic and antigenomic RNAs probably involves the consequences of S-HDAg binding. 2.5 Ribonucleoproteins Inside a delta virus particle the genomic RNA is bound to molecules of S-HDAg (Bichko et al. 1996; Dingle et al. 1998; Ryu et al. 1993). As discussed later in Sect. 3.1, this protein that is the only one encoded by HDV, exists in two main size classes. The 195-amino acid S-HDAg is essential for HDV replication. During replication RNA editing leads to a change in the amber stop codon of S-HDAg to tryptophan (see chapter by J.L. Casey, this volume). This so-called amber/w mutation, leads to the translation of a protein form that is 19 amino acids longer at the C terminus. The resultant 214 amino acid large delta protein (L-HDAg) is both an inhibitor of replication and is essential, along with the envelope proteins of HBV, for the assembly of new virus particles (Chang et al. 1991; Chao et al. 1990). However, additional findings indicate that it is only when this L-HDAg is isoprenylated that it can function to assist assembly (Glenn et al. 1992) or to act as an inhibitor of replication (Sato et al. 2004). In natural infections, the assembled HDV particles contain both forms of HDAg in a ribonucleoprotein structure with the HDV genomic RNA (Ryu et al. 1992). In virions there are about 70 molecules of S-HDAg bound per molecule of genomic RNA (Ryu et al. 1993). These interactions are facilitated by an RNA-binding domain that is shared by both forms of HDAg (Lee et al. 1993). Within a cell undergoing HDV genome replication the majority of the accumulated genomic and antigenomic RNAs exist in complexes with S-HDAg. This has been demonstrated by immunoaffinity procedures using antibody specific for S-HDAg; in contrast, the mrna for HDAg is not in such a complex (Nie, Chang, Taylor, unpublished). These findings are consistent with earlier reports that in vitro, S-HDAg can specifically recognize the rod-like folding of the genomic and antigenomic RNAs (Chao et al. 1991). A detailed comparison of the stoichiometry of S-HDAg per RNA has yet to be made for the intracellular ribonucleoprotein (RNP) relative to the virion RNP. Also, it will be important to determine the crystal structure of molecules of S-HDAg bound to a segment of HDV rod-like RNA.

15 Structure and Replication of Hepatitis Delta Virus RNA 7 3 RNA Replication 3.1 Roles of Delta Proteins The Essential Small Delta Protein For some time it has been clear that the 195-amino acid S-HDAg is essential for HDV replication (Chao et al. 1990). Many activities of this protein have since been reported, and as summarized in Table 1, many specific roles in the HDV life cycle have been proposed. However, not yet solved is how many of these proposed roles actually contribute to the essential nature of this protein during HDV replication. It should not be unexpected that this protein will have several roles Other Forms of Delta Protein While S-HDAg is essential for replication other forms of the protein arise during replication. The best characterized other form is L-HDAg. It arises as a consequence of RNA-editing at a specific site, nucleotide 1012, in one genome numbering scheme (Kuo et al. 1988a). This location corresponds to the middle of the amber termination codon of S-HDAg. The RNA editing is carried out by ADAR-1, an adenosine deaminase acting on RNA (Sect. 4.2). The L-HDAg contains a single cysteine, located four amino acids from its novel C terminus. This cysteine is isoprenylated in vivo, and plays an essential role in the ability of this L-HDAg to facilitate virus assembly (Chang et al. 1991; Table 1 Proposed roles of S-HDAg in HDV replication 1. Form an RNP that stabilizes the genome and antigenome (Chao et al. 1991; Lazinski and Taylor 1995) 2. Form an RNP that protects HDV RNAs against ADAR editing (Cheng et al. 2003) 3. Form an RNP that facilitates transport of the HDV genome to the nucleus (Xia et al. 1992) 4. ActasanRNAchaperonetoacceleratetheHDVribozymeactivities (Huang and Wu 1998; Jeng et al. 1996) 5. Act as a facilitator of processivity during RNA-directed RNA transcription (Yamaguchi et al. 2001)

16 8 J.M.Taylor Glenn et al. 1992). The L-HDAg does not support HDV genome replication and at least under certain conditions, acts as a dominant negative inhibitor of such replication (Chao et al. 1990; Sato et al. 2004). To consider HDV replication as being associated with just these two forms of HDAg is too simple. One has to factor in the consequences of posttranslational modifications, such as phosphorylation, acetylation, methylation, and isoprenylation (see chapter by W.-H Huang et al., this volume). In addition, it would seem that there are other RNA editing sites and there are certainly sites at which transcriptional errors occur (Sect. 4). Some of these sequence changes can lead to S-HDAg with altered sequence and functionality. Thus, once HDV replication is underway, there is a real heterogeneity in the amino acid sequence of those species, some of which will electrophoretically migrate the same as the prototypic S-HDAg or L-HDAg (Gudima et al. 2002). This heterogeneity is particularly true in situations where HDV replication is occurring in the absence of packaging followed by virus release and new rounds of infection; that is, when there is no selection for functional HDAg Replication with an Unchanging Separate Source of Small Delta Protein Because of some of the problems of delta protein heterogeneity described above in Sect , a recent study has established a cell system in which HDV replication occurs in the presence of an unchanging DNA-directed source of S-HDAg (Chang et al. 2005a). In this system an integrated cdna provides a single source, under tetracycline control, of S-HDAg. Added to these cells is an HDV RNA genome previously modified so that it can no longer express any S-HDAg. This new system provides a better approach to address important questions of HDV replication questions. Some applications will be described in subsequent Sections. 3.2 Enzymology of RNA-Directed Transcription Given that HDV genome replication involves RNA-directed RNA transcriptionandthats-hdagistoosmalltohavepolymeraseactivity,ithasbeenclear for some time that one or more host polymerases are required for HDV transcription. This being said, the characterization of such transcription leaves alottobedesired. Much evidence invokes the interpretation that replication involves the redirection of host DNA-directed RNA polymerase II, Pol II (Macnaughton et al. 2002; Modahl et al. 2000; Moraleda and Taylor 2001). However, a complication is that it has been interpreted that a second polymerase, one that is more resistant to alpha-amanitin than Pol II, might be

17 Structure and Replication of Hepatitis Delta Virus RNA 9 involved in the transcription of genomic RNA templates to produce antigenomic RNAs (Macnaughton et al. 2002; Modahl et al. 2000). In the absence of convincing experimental support it has been interpreted that RNA polymerase I is involved. This in turn has been incorporated into a highly speculative and complex rolling-circle model in which genomic RNA templates are sometimes transcribed by Pol II and other times by Pol I (Macnaughton et al. 2002). It is agreed that the transcription of antigenomic RNA templates into genomic RNA is sensitive to alpha-amanitin at levels consistent with the enzyme being Pol II. The accumulation of the HDV mrna species is similarly sensitive. Furthermore, there is the circumstantial evidence for Pol II, in that accumulation of this mrna is dependent upon poly(a) processing signals that in animal cells are only recognized by Pol II (Gudima et al. 2000; Hsieh and Taylor 1991; Nie et al. 2004). One would hope that stronger evidence obtained via robust experimental systems might be available for this important transcription question. The problem is that to date, no one has been able to obtain a reproducible and competent system for in vitro transcription of HDV RNA templates. One early in vitro study cannot be reproduced (Fu and Taylor 1993). Other in vitro studies achieve what is predominantly 3 -end addition to HDV RNAs, a process that might not be of biological relevance (Beard et al. 1996; Filipovska and Konarska 2000; Gudima et al. 2000). Why then has HDV transcription not been better characterized? The answermightbethatnoonehasbeenabletouseinvitrotranscriptionreactions to achieve credible initiation of HDV RNA transcripts. In part, this may be because in vitro transcription will probably make use of ribonucleoprotein structures rather than naked HDV RNAs. Currently some progress is being made by the application of immunoaffinity procedures following disruption of cells undergoing a burst of HDV replication (Nie, Chang, Taylor, unpublished), just as others have done for other RNA viruses (Qanungo et al. 2004; Waris et al. 2004). Following such selections, both genomic and antigenomic unit-length HDV RNAs can be found bound to S-HDAg, and a fraction of these are also bound to RNA polymerase II (Nie, Chang, Taylor, unpublished). However, such complexes will have to be proven as competent for in vitro transcription and they will have to be carefully characterized for all the host proteins present. 3.3 Initiation of Replication In a natural infection, a receptor-mediated interaction of the virus with the host cell leads to the viral RNP reaching the nucleus and then initiating RNA-

18 10 J. M. Taylor directed RNA transcription. Such infections have also been initiated using cultured primary hepatocytes but never with established cell lines. To initiate HDV replication in cell lines, many different strategies have been used. Some are listed in Table 2. Several important points need to be made about these systems: (1) S-HDAg has to be present at or soon after the transfection; (2) the in vitro RNAs and the in vivo DNA-directed transcripts are linear RNAs, not circular, as in the virions; (3) in some cases, even the total nucleic acid extracted from a cell undergoing replication, can be transfected into new cells and is sufficient to initiate genome replication; (4) for the method described in Sect , a low level of replication is initiated by transfecting HDV RNA into cells continuously expressing only a low level of S-HDAg. When these cells are induced by addition of TET to express large amounts of S-HDAg, there can be a rapid burst of replication; presumably this is because the RNA templates are already circular species resident within the cells. It has to be realized that in many transfection studies the HDV replication is initiated by nucleic acid species that are different from the circular genomic RNA present in the virions that initiate a natural infection. For some of these transfections the initial HDV RNA templates were greater than unit-length tandem multimers. It was considered that since each unit-length RNA, be it genomic or antigenomic, will contain a functional ribozyme, then initiation of replication might need an initial conversion of the RNA to unit-length linear and then to a circle, presuming that such circles might be the preferred template for RNA-directed transcription. While this may be true, recent studies show that linear RNAs, even unit-length RNAs, do not necessarily have to Table 2 Components used for transfections that can initiate HDV genome replication in cell lines 1. HDV virions (Bichko et al. 1994) 2. HDV RNP from virions (Bichko et al. 1994) 3. HDV cdna in expression vectors (Kuo et al. 1989) 4. HDV RNA transcribed in vitro, into cells expressing S-HDAg (Glenn et al. 1990) 5. HDV RNA transcribed in vitro, pre-mixed with recombinant small S-HDAg (Dingle et al. 1998a) 6. HDV RNA transcribed in vitro, together with in vitro transcribed mrna that can be translated into S-HDAg (Modahl and Lai 1998) 7. As in 4, but using total RNA extracted from cells in which replication was occurring (Gudima et al. 2004)

19 Structure and Replication of Hepatitis Delta Virus RNA 11 be converted to circles before they can be transcribed to initiate replication (Chang and Taylor 2002). This result has made us aware that even after the initiation a natural infection, some of the nascent HDV RNA species that are noncircular and possibly of unit length or greater, might actually function as templates for transcription. To assess this contribution, a competition assay between RNA circles and linear HDV RNAs was able to show that unit-length circles are about 15 times better at initiation than linear RNAs. Also, genomic and antigenomic RNAs are of equal efficiency in initiation (Gudima et al. 2004). Many questions remain regarding the initiation of replication in a natural infection. Does S-HDAg function directly in this process and if so, how? What site(s) on the genomic RNA does the host polymerase recognize to facilitate such initiation? From what sites does the initiation actually take place? Are we correct in presuming that the 5 -end of the mrna arose via sitespecific primer-independent initiation? Later, when new antigenomic RNAs are produced, again what does the host polymerase bind to and where is transcription initiated from? 3.4 Double Rolling-Circle Model For some time, attempts have been made to create models for the replication of HDV RNAs (Flint et al. 2004; Gerin et al. 2001; Macnaughton et al. 2002; Taylor 1990). Most of these attempts have borrowed from the concept of a rolling-circle model of replication. This idea was applied previously to the replication of RNAs of the plant viroids (Branch and Robertson 1984), with which HDV has numerous similarities, as discussed in detail elsewhere (Taylor 1999). The viroid RNAs differ from HDV RNAs in that they are several times smaller and are noncoding. This, together with the fact that plant viroid RNAs are never assembled into virus-like particles, means that the concept of a viroid genome has to be different for HDV. For some viroids RNA-directed RNA replication leads to the accumulation of unit-length RNA circles of both polarities. In contrast, for most viroids, circles of only one polarity can be found. For the complementary strand the RNA template is a linear RNA multimer of unit-length. It is considered that these linear RNAs act as templates for multimeric transcripts that can be processed to unit-length circular RNAs. Thus, for these two classes of viroids the replication models are described as double- and single-rolling circle models, respectively. For HDV, since both the genome and antigenome exist as unit-length circles, it was quickly extrapolated from the viroids, that the replication would

20 12 J. M. Taylor Table 3 Some revisions to rolling-circle model of HDV genome replication 1. Nascent antigenomic transcripts can be ribozyme-cleaved independent of polyadenylation (Nie et al. 2004) 2. Nascent antigenomic transcripts can be polyadenylated independent of ribozyme cleavage (Nie et al. 2004) 3. Nascent antigenomic transcripts, undergo polyadenylation or ribozyme-cleavage as largely alternative processing events (Nie et al. 2004) 4. While circular forms of genomic and antigenomic RNA are preferred templates for RNA-directed transcription, linear forms can also act (Gudima et al. 2004) be a double-rolling circle model. Furthermore, this basic model had to be adapted, to allow for the fact that a polyadenylated mrna was also being produced. This model, with time, has required a number of additional modifications,someofwhicharelistedintable3. Without resorting to a diagram, Table 4 attempts to describe the series of events that could be considered as essential steps in HDV genome replication. At the same time it should be clear from this review, and maybe from the other reviews in this volume, that for many of these steps we have yet to obtain actual experimental evidence. 3.5 Template-Switching, Reconstitution, and Recombination From studies of HDV replication as initiated by the transfection of linear HDV RNAs, it is clear that template-switching can occur during transcription of the transfected RNA. The best evidence for this is that when cells are transfected with linear RNAs that are one or two nucleotides less than unitlength, replication can be initiated but there are specific deletions and even nontemplated additions, on the RNAs that replicate (Chang and Taylor 2002). With this knowledge that template-switching can occur, experiments were undertaken in which the transfected RNA was replaced by two RNAs, each of which was significantly less than unit-length, but which together provided representation of the whole genome. Following transfection with such RNAs, genome replication was detected, consistent with reconstitution of the HDV genome (Gudima et al. 2005). Such reconstitution was only achieved when the two RNA templates were pre-associated before the transfection. In fact, all of the available data concerning HDV template-switching is consistent with the role of the rod-like folding as a facilitator. That is, the inter-molecular association achieved prior to transcription depends on utilization of base-pairings normally considered to be part of the intra-molecular rod-like

21 Structure and Replication of Hepatitis Delta Virus RNA 13 Table 4 Some key steps in HDV RNA-directed RNA transcription during a natural infection 1. Following virus attachment and entry the genomic RNA, as a ribonucleoprotein (RNP) complex with S-HDAg, is transported to the nucleus 2. This RNP is able to redirect either an inactive Pol II complex or one already active in DNA-directed RNA transcription 3. One of the sites for initiation of transcription on the genomic RNA corresponds to position 1630, the 5 -end of the mrna 4. Following initiation, elongation using the circular RNA template, can produce antigenomic transcripts that are greater than unit length 5. Such nascent antigenomic RNA transcripts can be processed either to mrna or to unit-length circular antigenomic RNA, both of which are relatively much more stable than the nascent transcript 6. The mrna species are transported to the cytoplasm and the translation product, new S-HDAg, returns to the nucleus to support more RNA-directed RNA transcription. It may also alter the balance of processing in step 5 7. Transcription of new antigenomic RNA templates occurs, with the nascent transcripts being processed to form new unit-length genomic RNAs 8. ADAR editing can occur on all nascent genomic and antigenomic RNAs, and/or on processed unit-length RNAs. Essential to the life cycle is that some nascent antigenomic RNA and/or processed unit-length antigenomic RNAs, become a target for specific ADAR-editing at position 1012, leading to the production and translation of mrnas encoding L-HDAg 9. After editing and maybe other sequence changes, translation produces altered forms of HDAg, especially of L-HDAg, that fail to support or even act as inhibitors of further RNA-directed transcription. In addition, L-HDAg, after isoprenylation, can complex with genomic RNA that in turn can interact with the envelope proteins of the helper virus HBV, if present, to achieve assembly and release of new virus particles folding. Subsequently, it is considered that these new base-pairings within the RNA hybrid template force pauses in transcription at locations which allow template-switching to occur and achieve reconstitution of replication competent HDV RNA. It might be argued that the above examples of template-switching, although achieved within a cell rather than in vitro, are more relevant to the question of what an RNA polymerase can do, rather than to how an HDV genome is normally replicated. However, such studies are relevant to the question of whether there can be recombination between HDV RNA genomes. Some data from examination of patients infected with two different HDV genotypes have been interpreted as evidence for inter-molecular recombina-

22 14 J. M. Taylor tion (Wang and Chao 2005; Wu et al. 1999). Also, it has been asserted that recombination can be achieved in cells transfected with two different HDV genotypes (Wang and Chao 2005). However, these data are not yet convincing and it is known that other attempts involving transfected RNAs have proven negative (Gudima et al. 2005). 3.6 Inhibition of Replication Resistance to Interferons InterferonsareoftenusedaspartoftreatmenttherapiesforHDV(asdiscussed in chapters by J.S. Glenn and J.L. Casey and J.L. Gerin, this volume). However, when applied to HDV replication as it occurs in cultured cells in the absence of HBV, no such inhibition has been detected with interferons α or γ (Chang et al. 2006; Ilan et al. 1992; McNair et al. 1993). In apparent contrast, a fraction of patient therapies are successful with high dose interferon treatments (Kleiner et al. 1993; Lau et al. 1999). However, it might be that these treatments are interfering with HDV indirectly, by inhibiting the helper virus HBV Sensitivity to Ribavirin Someyearsagoitwasshownthatribavirincouldblockthereplicationof HDV in primary woodchuck hepatocytes (Choi et al. 1989; Rasshofer et al. 1991). Recent studies confirm that this occurs in cell lines undergoing HDV replication (Chang et al. 2006). Moreover, this inhibition can be achieved with 30 µm ribavirin, a dose low enough to avoid cell toxicity. In contrast to this, others have cited that ribavirin treatments for HDV-infected patients are not effective (Hoofnagle 1998). However, given that ribavirin treatment is demonstrably selective for HDV replication in cultured cells, further studies in patients may be warranted. What might seem a possible hindrance to patient studies is that a side effect of ribavirin treatment can be anemia (Galban- Garcia et al. 2000). However, such side effects can be controlled, as judged by the current acceptance of ribavirin (combined with pegylated interferon) as partofatreatmentforchronichepatitiscvirusinfection.also,ribavirinmight be replaced with viramidine, an immediate precursor to ribavirin, a drug that more specifically targets the liver and should have fewer side effects (Lin et al. 2003). This drug, when applied to cultured cell lines at an appropriate concentration, can also specifically inhibit HDV genome replication (Chang et al. 2006).

23 Structure and Replication of Hepatitis Delta Virus RNA Sensitivity to sirna Small interfering RNAs (sirna) are short double-stranded RNAs of about 21 base pairs. They have offered great promise as ways to interfere with virus replication and therapies in humans are already underway. Many RNA viruses have been tested and found susceptible to sirna attack (Bitko and Barik 2001; Coburn and Cullen 2002; Ge et al. 2003; Gitlin et al. 2002). Consistent with this, the replication of HDV in cultured cells can be inhibited by transfection of appropriate sirna species (Chang and Taylor 2003). A caveat here is that only sirna targeted against the HDV mrna produce such inhibition. Those targeted against other regions on the genome or antigenome do not block replication. One possible reason for resistance is that the genomic and antigenomic RNAs are located in the nucleus, away from the RISC complex that is considered to mediate sirna action. Another possibility is that the binding of S-HDAg to these RNAs confers resistance to sirna-mediated degradation Resistance to Dicer Dicer is an enzyme present in animal cells that can act on RNA species that have 100% base pairing, to release sirna species (He and Hannon 2004). Dicer also plays a role in the cleavage of micrornas. These, like sirna, are of about 21 nucleotides in length. They are derived from regions of RNA transcripts that have significant levels of intra-molecular base pairing. Such precursors are first cleaved in the nucleus by an enzyme known as drosha (Lee et al. 2003). The fragments produced are frequently RNA hairpins of about 70 nucleotides, with extensive but <100% base pairing. These precursors are cleaved further in the cytoplasm by dicer to release the micrornas. The similarity between such precursors and the rod-like folding of HDV genomic and antigenomic RNAs is striking and leads to the question of whether HDV RNAs are also cleaved by dicer. Furthermore, there were reports that during the replication of two plant viroids, each with rod-like RNA genomes, that sirna were detected (Itaya et al. 2001; Martinez De Alba et al. 2002). However, when an examination for sirna was made for HDV RNAs under various conditions of replication, no sirna couldbedetectednorcouldanysirnabegeneratedwheninvitrotranscribed HDV RNA species were subjected to recombinant dicer (Chang et al. 2003). 3.7 Cytopathic Effect of Replication From histological examinations of liver at the peak of natural and experimental HDV infections, whether fulminant or nonfulminant, it has been concluded