Hepatitis D virus virions are roughly spherical particles (36–43 nm) (3, 101) containing a ribonucleic core-like structure surrounded by HBV envelope proteins and host lipids. The core-like structure (19 nm) is composed of one HDV genomic RNA complexed with about 70 molecules of HDAg in its large and small forms (102). HDAg form dimers through an antiparallel coiled coil. Dimers then interact to form octamers, which are arranged in 50 Å rings into a unique structure (103). The HDAg is thus not exposed at the virion outer surface. The envelope is composed of about 100 copies of the three envelope proteins of HBV: the small HbsAg (S-HBsAg), the middle (M-HBsAg) and the large (L-HBsAg). The relative proportion of these proteins, 95:5:1, is more similar to that found in the 22 nm, HBsAg-positive empty particles of HBV than that of the Dane particle (55). It has also been demonstrated that S-HBsAg is sufficient for particle assembly, that L-HBsAg is necessary for infectivity (104) and that, finally, M-HBsAg is not essential for assembly or infectivity (105).
L-HDAg (but not S-HDAg) and HBsAg are necessary and sufficient to form particles, which obviously are not infectious unless also HDV RNA is present (106). S-HDAg increases packaging efficiency (106, 107). The antigenomic strand has never been found in the viral particle.
Hepatitis D virus RNA
Three different RNAs accumulate during HDV replication: the genome (∼300 000 copies), the antigenome (∼30 000 copies) and the mRNA (∼600 copies).
The genome is a circular negative single-strand RNA composed of 1672–1697 nucleotides, depending on the strain (108). It is the smallest and the only circular RNA among animal viruses, this structural characteristic and mode of replication being otherwise only observed in plant viroids and virusoids. Because of the presence of 74% internal base pairing, this molecule has the ability to fold on itself as an unbranched, double-stranded, rod-like structure (4, 109). Nucleotide number one has been arbitrarily chosen using a unique Hind III restriction site (110). The genome contains a ribozyme domain spanning nucleotides 680–780 and a putative promoter site for the HDAg mRNA (111).
The antigenome is the exact complement of the genome, as replication occurs through RNA-directed RNA synthesis without any DNA intermediates (112). The antigenome contains a unique open reading frame (ORF) coding for the HDAg and, as the genome, a ribozyme domain (113, 114).
Finally, the mRNA directs the synthesis of HDAg. It is an 800-nucleotide linear RNA of the same polarity as the antigenome, which bears a 5′ cap structure and a 3′-poly(A) tail (115–117). The mRNA, at variance with the genome, is associated with polysomes during replication and can be translated in vitro (118). This molecule is unstable and is synthesized throughout replication (116) (Fig. 3).
Hepatitis D antigen
The only protein that is expressed by HDV is HDAg. A unique ORF, located on the antigenome, leads to the synthesis of two different forms of HDAg during the HDV cycle: S-HDAg (195 amino acids; 24 kDa) and L-HDAg, which has 19 additional amino acids at its C-terminus (214 amino acids; 27 kDa). The large form appears later, when some antigenomes are post-transcriptionnally edited by an enzyme, the adenosine deaminase acting on RNA-1 (ADAR-1) (119, 120). The adenosine at the amber/W site (amino acid position 1015) is deaminated into an inosine (UAG becomes UIG), and then paired to a cytosine during the next replication cycle (AUC becomes ACC). When this modified genome is transcribed into mRNA to produce an antigen, the stop codon (UAG) is replaced by a tryptophan (UGG) (118, 119, 121) (Fig. 4). The synthesis of L-HDAg then follows, reaching the next stop codon, which is located 19 codons downstream.
S-HDAg is necessary for the initiation of genome replication (37), while the L-HDAg behaves as a dominant negative inhibitor (122, 123) and is essential for the assembly of the particle (106, 124).
HDAg binds to the RNA, an ability that is perhaps facilitated by the rod-like structure (125). HDAg may directly stimulate transcription elongation via the replacement of the negative elongation factor, a transcription repressor bound to the RNA polymerase II (RNAP II) (126).
In the absence of HBsAg, both L-HDAg and S-HDAg localize in the nuclei as they bear nuclear localization signals spanning amino acid residues 35–88 from the N-terminus (127, 128). In the presence of HBsAg, L-HDAg relocalizes to the cytoplasm (129). L-HDAg is indeed a nucleocytoplasmic shuttling protein as it also possesses a nuclear export signal in its C-terminus (130).
Several post-translational modifications of HDAg have been reported. L-HDAg contains a terminal CXXX box, which is a substrate for isoprenylation, a modification that enhances its replication inhibitory effect (131) and is necessary for viral particle formation. The prenylated L-HDAg is a lipophilic molecule that mediates the direct binding between the L-HDAg and the HBV envelope proteins (132).
Both HDAg forms are phosphorylated, although L-HDAg is six times more phosphorylated than S-HDAg (133). S-HDAg is phosphorylated at serine and threonine residues, while L-HDAg is phosphorylated at serine residues only (134, 135). The major phosphorylation site is the serine 177. The protein kinase R has been shown to modulate HDV replication by phosphorylating S-HDAg (136). Phosphorylation of the serine 2 of S-HDAg increases HDV replication (137). HDAg phosphorylation is important for RNA replication, probably because it mediates its RNA-binding activity (137, 138).
The methylation of S-HDAg, which is observed in vitro and in vivo, is located in the RNA-binding domain. Mutations of this domain or use of a methylation inhibitor result in an inhibition of HDV RNA replication: HDAg loses the ability to form speckled structures in the nucleus and localizes in the cytoplasm (139).
Acetylation is another post-translational mechanism shown to modulate HDV replication as it regulates the nuclear localization of HDAg. A substitution of the lysine 72 by an alanine decreases the accumulation of viral RNA and induces an earlier appearance of L-HDAg (140).
Sumoylation is a newly described type of post-translational HDAg modification. S-HDAg is a small ubiquitin-like modifier 1 (SUMO1) target protein through multiple lysine residues. Experiments performed with S-HDAg fused to SUMO1 protein showed an increased HDV genomic RNA and mRNA synthesis, but no influence on antigenomic RNA synthesis (141) (Fig. 5).
Figure 5. Localization of HDAg functional domains: RNA-binding domain, coiled-coil sequence, nuclear localization sequence and virus assembly signal. Sites of post-translational modifications: phosphorylation (P), methylation (M), acetylation (A) and isoprenylation (I). Adapted from (186).
Download figure to PowerPoint
Virus attachment, entry, assembly and release
At of the time of writing, the receptor of HDV and its entry mechanism were unknown. Some information is nevertheless available. HDV and HBV share the same envelope proteins; we would thus expect the same or very similar mechanisms of attachment and entry for both viruses. However, HBV becomes non-infectious (at variance with HDV) when the virion is packaged with S-, M- and L-HBsAg lacking the N-linked carbohydrates (142). However, some similarities exist: residues 5–20 of the pre-S1 domain in the L-HBsAg are required for the entry of both viruses. In fact, entry can be inhibited by synthetic acylated (143) and myristylated (144, 145) peptides encompassing the first 50 amino acids of the pre-S1 domain. Entry appears to be preceded by attachment to the carbohydrate side chains of hepatocyte-associated heparan sulphate proteoglycans (146) as suramin inhibits in vitro HDV infection (147).
One of the classical routes of viral entry that could also be implicated is the clathrin-mediated endocytic pathway (129). Viruses such as the influenza virus (148), the reovirus (149) or the vesicular stomatitis virus (VSV) (150) are endocytosed together with their receptor into endosomes. HDAg has been identified as a clathrin adaptator-like protein as it specifically interacts with the clathrin heavy chain (CHC) (151). Clathrin is also implicated in the exocytosis mechanisms of some viruses such as the VSV or the HIV type 1. Interestingly, HDV assembly is reduced after CHC downregulation whereas HBV is not, suggesting that even though both viruses share the same envelope proteins, their assembly and release mechanisms are different (152). The assembly efficiency is, in addition, different between genotypes and correlate with the ability to interact with CHC.
Hepatitis D virus genotype 1 has an assembly efficiency that is higher than genotypes 2 and 3. A recent work has pointed out the implication of the amino acid in position 205 in virion release. Substitution of the proline by an arginine or an alanine in genotype 1 significantly decreased the secretion of viral particles. The reverse substitution in genotypes 2 or 3 increased the assembly efficiency of HDAg (153).
Hepatitis D virus RNA replication
As HDV does not possess its own RNA polymerase, the virion uses the human transcriptional machinery for its replication. The implication of the RNAP II is now well established. RNAP II binds to the HDV genome and antigenome (154), the HDV replication is sensitive to α-amanitin (155) and the mRNA coding for the HDAg possesses a 5′ methylguanine cap and a poly(A) tail (115, 156). Resistance to α-amanitin, however, does not occur for the transcription of the antigenome, suggesting an involvement of RNAP I (157). Other evidences include the interaction between components of the SL1 transcription factor for RNAP I and S-HDAg (158), inhibition of the antigenome synthesis by anti-α-SL1 antibodies (158) and direct interaction between RNAP I and HDV RNA (159). Recent data suggest an implication of the third polymerase, the RNAP III, which also binds to both genomic and antigenomic HDV RNA (159).
Transcription occurs in the nucleus, where HDV RNA is brought into by the HDAg (160) and is activated by S-HDAg binding to HDV RNA. Replication occurs without any help coming from HBV and without any DNA intermediate. Replication proceeds via a double-rolling circle process, a mechanism proposed for the replication of plant viroids and small-satellite RNAs of some plant viruses (161). The transcription proceeds for more than one genome length, going twice through the ribozyme cleavage site. Then the antigenomic transcript undergoes an autocatalytic cleavage between nucleotides 688 and 689 via the antigenomic ribozyme, is folded into a rod-like structure and is ligated by cellular ligases (162) to form a circular template. Production of the genomic RNA from the antigenome occurs through the same mechanism; then the genomic RNA is either incorporated into new viral particles or used again as template for antigenomic RNA synthesis. The ratio between genomic and antigenomic RNA is asymmetric, as the genome is fifteen times more abundant than the antigenome (112). The mechanism of regulation is unknown but it has been reported that HDV RNA mutations could suppress genomic strand transcription, without affecting antigenomic strand synthesis, suggesting two separate means of regulation (163). The mRNA is transcribed from the genome, thus using the same template as for the antigenomic RNA. The transcription of these two elements nevertheless starts at different initiation sites (116), using different polymerases and in different subnuclear compartments. The antigenome is transcribed into the nucleolus, whereas the mRNA is transcribed into the nucleoplasm, like the genome (Fig. 6).