Biotecnologia Aplicada Elfos Scientiae ISSN: 0684-4551 Vol. 17, Num. 2, 2000, pp. 122

Biotecnología Aplicada 2000;17:122

Modelling of Hepatitis C Proteins

Asutosh Yagnik, Armin Lahm, Anna Tramontano

IRBM P. Angeletti Via Pontina Km 30.600 - I00040 Pomezia (Rome).
URL: http://www.irbm.it; E-mail: Tramontano@irbm.it

Code number: ba00035

Introduction

Hepatitis C virus (HCV) has been identified as the causal agent of most transfusion-associated non-A non-B hepatitis infections. It afflicts millions of people world wide and its infection has a chronicity rate of about 70% [1]. During the last 10 years, tremendous progress has been achieved in our understanding of the biology of hepatitis C virus, but neither a vaccine nor an effective antiviral therapeutical agent have been yet developed.

HCV has a single -stranded RNA genome of about 9.6 kb in length which encodes a precursor polyprotein composed of 3010-3030 residues. The precursor polyprotein comprises the structural proteins (C, E1, E2) and the nonstructural proteins NS2-NS5B, and is cleaved into individual proteins. The proteolytic cleavage at the NS3/NS4A, NS4A/NS4B, NS4B/NS5A, NS5A/NS5B sites is mediated by a chymotrypsin-like viral serine proteinase encoded within the NS3 protein. NS5A has been shown to be the RNA -dependent RNA polymerase. Several experimental evidences suggest that HCV E2 is a fundamental candidate antigen for a vaccine against hepatitis C virus. NS3 protease and NS5B are important targets for the development of antivirals.

Models of all these three proteins were built using different techniques and have been very useful in suggesting critical experiments. In two cases (NS3 protease and NS5B), their general features have been confirmed by X-ray crystallography.

Materials and Methods

CLUSTALW was used for multiple sequence alignments, PHD for secondary structure prediction, TOPITS, THRADER2 and ProCyon for fold recognition. Insight II was used for model building. The stereochemical quality of the models was examined using PROCHECK. The description and references for these software packages can be found at the URL: http://www.irbm.it/irbm-course97.

Results and Discussion

Since the E2 protein sequence does not share significant sequence homology with any known protein, we applied several protein secondary structure prediction and fold recognition techniques. This led to the construction of a model based on the E protein of tick-borne encephalitis virus (1SVI3). Mapping all available experimental data onto this structure allowed the binding interactions between E2 and its proposed cellular receptor CD81 to be localised, as well as a rough model for the tertiary and quaternary structure of the envelope glycoproteins E1 and E2 to be proposed [Yagnik A, Lahm, A T, unpublished results].

The NS3 serine proteinase domain shows only limited homology to cellular serine proteinases. Despite this low sequence conservation, we built a homology model of this domain that has helped us in directing experimental work, even in the absence of detailed structural information [2]. The three-dimensional structure of the HCV NS3 proteinase domain has now been solved by X-ray crystallography [3] and revealed that we correctly predicted the general topology, as well as several unique structural features of the enzyme, including the positions of the residues that determine the shape of the S1 substrate binding pocket and the presence of a tetrahedral metal binding site.

An inital partial model of the NS5B RNA-dependent polymerase was based on the coordinates of the poliovirus 3Dpol X-ray structure. For the sequence alignment we took advantage of the presence of conserved sequence motifs in RNA polymerases and of the secondary structure prediction for HCV NS5B. Superposition of the crystal structures of T7 DNA polymerase and of HIV reverse transcriptase, both in complex with a primer/template substrate, allowed us to identify NS5B residues that might be involved in substrate recognition and/or catalysis. Based on the now available NS5B crystal structure, we know that most of the palm domain and the fingers domain topology was correctly predicted. The model also correctly predicted the elements forming the proposed active site (A Lahm, unpublished results).

In conclusion, our results demonstrate that even models based on very low sequence homology, and therefore necessarily approximate, can be extremely useful in the drug discovery process.

References

1. Clarke B. Molecular virology of hepatitis C virus. J Gen Vir 1997; 78:2397–410.

2. Pizzi E, Tramontano A, et al. Molecular model of the specificity pocket of the hepatitis C virus protease: implications for substrate recognition. Proc Natl Acad Sci USA 1994;91: 888–92.

3. Yan Y, Li Y, Munshi S, et al. Complex of NS3 protease and NS4A peptide of BK strain hepatitis C virus: a 2.2 Å resolution structure in a hexagonal crystal form. Protein Sci 1998;7:837–47.

Papers from Biotecnología Habana`99 Congress.
November 28-December 3, 1999.