Interfering with capsid formation: A practicable antiviral strategy against hepatitis B virus?


Deres K, Schroder CH, Paessens A, Goldmann S, Hacker HJ, Weber O, Kramer T, et al. Inhibition of hepatitis B virus replication by drug-induced depletion of nucleocapsids. Science 2003; 299: 893896. (Reprinted with permission from the American Association for the Advancement of Science.)

Michael Kann M.D.*, * Institute of Medical Virology University of Giessen Giessen, Germany.


Chronic hepatitis B virus (HBV) infection is a major cause of liver disease. Only interferon-α and the nucleosidic inhibitors of the viral polymerase, 3TC and adefovir, are approved for therapy. However, these therapies are limited by the side effects of interferon and the substantial resistance of the virus to nucleosidic inhibitors. Potent new antiviral compounds suitable for monotherapy or combination therapy are highly desired. We describe nonnucleosidic inhibitors of HBV nucleocapsid maturation that possess in vitro and in vivo antiviral activity. These inhibitors have potential for future therapeutic regimens to combat chronic HBV infection.


Persistent hepatitis B is still the major cause of liver cell carcinoma worldwide. Public perception of first-world hepatitis is dominated by hepatitis C virus (HCV), while infection with the hepatitis B virus (HBV) is thought to be predominantly restricted to Asia and Africa. However, even in the first world, despite the availability of a safe and effective vaccine and various therapeutic modalities, prevalence of hepatitis B is still between 0.2 and 0.5 %.1

HBV shows some homology to retroviruses in that it contains a DNA genome but replicates by way of an RNA intermediate known as a pregenome (PG). The PG is synthesized within the nucleus of the infected hepatocyte by cellular RNA polymerase II2 using the viral DNA genome as the template (Fig. 1A; for review, see Seeger and Mason3). After being exported into the cytoplasm, the PG encodes for the viral capsid protein (also termed core protein) and the viral polymerase (pol) (Fig. 1B). Next, pol specifically interacts with the PG4 while the capsid proteins interact with each other, resulting in spontaneous assembly to capsids (core particles).5 Capsid assembly is combined with the encapsidation of a pol–PG complex6 and a cellular protein kinase,7 which has not yet been unequivocally identified (Fig. 1C). Inside the lumen of the capsids, pol first converts the PG into minus-strand DNA by reverse transcription, followed by synthesis of an incomplete plus-strand DNA (Fig. 1D). The DNA-containing capsids can either mediate the transport of the viral genome into the nucleus8 (Fig. 1E) or they can be enveloped with the viral surface proteins, resulting in secretion of progeny virus9 (Fig. 1F). At least the nuclear transport involves phosphorylation of the capsid proteins.10, 11

Figure 1.

Intracellular life cycle of the hepatitis B virus. After infection, the covalently closed circular DNA form of the virus within the nucleus encodes for the viral messenger RNAs. (A) Multiplication of genomic information is achieved by synthesis of the pregenomic messenger RNA (PG). (B) After being exported to the cytoplasm, the PG serves as the template for translation of the viral pol and the capsid proteins. (C) PG, pol, and a cellular protein kinase assemble to an immature capsid. (D) Inside the capsid, pol converts the encapsidated RNA into a partially double-stranded (pds) DNA genome. (E) These DNA-containing capsids can be transported through the nuclear pore complex (NPC) into the nucleus, which leads to amplification of the nuclear viral DNA early in infection. (F) Later, when viral surface proteins are synthesized, these capsids can be enveloped and secreted as progeny virus. The HAPs inhibit the assembly of the capsids (C). Thus all steps downstream should be inhibited, as it was shown for genome maturation (D). According to the HBV life cycle, HAP treatment should additionally block nuclear transport of the viral genome (E) and virus secretion (F).

HBV—in contrast to HIV and HCV, which, among others, harbor an essential protease—shows only the polymerase as its genuine enzymatic activity. Thus, nucleos(t)ide analogues (NAs), which affect either the reverse transcriptase or the DNA-dependent polymerase activity of pol, are successfully used to interrupt HBV multiplication. A similar reduction of the viral load can be achieved by using interferon-α, a treatment that has been established for more than 20 years. Both treatments, however, show only limited success. In only 30% of all patients—approximately—can sustained response be achieved if virologic parameters such as a significant reduction of viral load and HBeAg elimination and/or disease-related parameters such as normalized alanine aminotransferase levels and improved liver histology are chosen as markers. Although successful interferon-α therapy improves survival and quality of life, virus eradication occurs only in rare cases.12, 13 The same is true for NAs, although long-term experience with NAs is very limited. In addition, resistance is likely to occur with all NAs, as has been documented for lamivudine and adefovir dipivoxil.

Various experimental strategies have therefore been investigated to block the hepadnaviral life cycle at points other than viral polymerase. In addition to numerous investigations using antisense oligonucleotides or ribozymes, these approaches include the generation of capsid mutants that contain nuclease digesting the encapsidated PG14 or dominant negative capsid protein mutants,15, 16 which are unable to support PG packaging and genome maturation, as well as peptides interfering with capsid assembly.17 However, all of these studies are still at the very early experimental level.

Based on a previous work of this group18 in which the authors identified the nonnucleosidic compound Bay 41-4109 to be effective in the HBV transgenic mouse model, Deres and colleagues19 next analyzed the mode of action. In addition to Bay 41-4109, they included the congeners Bay 38-7690 and Bay 39-5493 in their investigation and compared them with the established NA lamivudine. They found that all three nonnucleosidic inhibitors, also referred to as heteroaryldihydropyrimidines (HAPs), showed a dramatic inhibition of HBV replication in HBV-producing cell lines. By titrating the HAPs to the stably HBV-producing cell line HepG.2.2.1520 and measuring the amount of HBV DNA, the authors showed that the HAPs had a twofold- to tenfold-lower IC50 than lamivudine (20–150 vs. 300 nM), which is currently the most frequently used NA in hepatitis B treatment. Only the (−)-enantiomers and not the (+)-enantiomers showed this antiviral effect, implying a very specific interaction between drug and target.

Exploring the mode of inhibition, they confirmed that the decreased number of synthesized HBV genomes corresponded with a decrease in viral polymerase activity, as would be expected. In contrast, the amount of viral RNAs remained unchanged. Up to this point, the results could be explained by classical NA activity; however, in contrast to NA, HAP treatment decreased the number of capsids and the amount of capsid protein in stably HBV expressing cells. This observation implied an inhibition of translation, yet there was no inhibition of capsid protein synthesis. Pulse chase experiments showed that the amount of newly synthesized capsid protein was similar in Bay 39-5493–treated and untreated cells. In addition, this assay showed that unassembled capsid proteins have a much shorter half-life (only 3 hours) than the assembled particles, which was calculated to be more than 24 hours. Although no answer is provided in this article, it appears that the difference in half-life is not caused by protease-mediated degradation of the non-assembled capsid protein, which may have been induced by HAPs treatment. More likely, the different turnovers are intrinsic properties of assembled capsid and unassembled capsid protein; this explains why intranuclear capsids show such a strong accumulation in persistently HBV-infected human liver cells. A conclusive explanation for these observations is that the HAPs prevented capsid formation and that the nonassembled capsid proteins were subjected to rapid degradation.

An obvious next step was to ask whether or not the HAPs directly interact with the capsid protein. To address this question, the authors expressed the core protein in Escherichia coli and performed in vitro capsid assembly assays. Using this experimental strategy, they could show that the block of capsid formation correlated with reversible in vitro binding of the HAPs to the capsid protein.

In conclusion, Deres et al. showed consistent and exciting evidence that the inhibition of capsid assembly becomes a real and reachable target in HBV drug treatment, although it is difficult to fully assess whether or not all three HAPs have the same effect given the limited information provided in the article.

The major question remains as to whether or not the HAPs will “do their job” in an in vivo model of hepatitis B and ultimately in patients with hepatitis B. The previous work of Weber and colleagues18 has already showed that at least Bay 41-4109 works in HBV-transgenic mice; however, in these mice no reduction of viral covalently closed circular DNA—the key problem in HBV treatment—can be studied. Since the HAPs (or at least Bay 39-5493) were shown not to interact with the duck hepatitis B virus (DHBV) capsid, it is unlikely that they would work in the commonly used DHBV duck model system. However, the DHBV capsid is not closely related to the HBV capsid. Consequently, one may ask why the authors have not tested the woodchuck hepatitis virus capsid, which shows a much closer homology to the HBV capsid sequence and would allow evaluation in an experimental animal system.