HIVdb: A Database of the Structures of Human Immunodeficiency Virus Protease

Human immunodeficiency virus protease (HIV PR) was only discovered as encoded in the HIV genome in 1985, but soon thereafter, this virus-specific enzyme was identified as a crucial target for designing drugs against acquired human immunodeficiency syndrome (AIDS). With six such drugs approved to date since 1995 by the United States Food and Drug Administration and several others in current clinical trials, the initial promise of rational drug design that originally seemed to be overly optimistic has been fulfilled beyond expectations. Introduction of PR inhibitors has changed the clinical outcome of AIDS and transformed an invariably fatal disease into a manageable one, although serious side effects and the development of resistance are still major unsolved problems. Crystal structures of HIV PR were first reported in 1989 and their availability had a major role in the process of drug development (although it is clear that the structures provided only a small fragment of information necessary for drug design). All pharmaceutical companies that succeeded in creating PR inhibitor drugs and many that either discontinued the efforts or are still conducting such research have been involved in solving numerous crystal structures of HIV PR. Most of these structures were of complexes of the enzyme with inhibitors that were either potential or actual drugs, or intermediate compounds useful in drug-design efforts. In addition, many academic laboratories joined the field and solved crystal and nuclear magnetic resonance structures of the complexes of HIV-1, HIV-2, and simian immunodeficiency virus (SIV) PRs with many diverse inhibitors. Because many variants of these proteins are known, attributable either to natural variation in the viral genomes or to their rapid mutation as a result of drug resistance, numerous mutant structures were also solved in pharmaceutical and academic laboratories. Although the total number of structures of these enzymes is not known, it can be estimated at many hundreds, making HIV PR the most widely studied enzyme in the history of protein crystallography. During the last 13 years, many structures of HIV PR were published and as many as approximately 150 have been deposited at the Protein Data Bank (PDB). However, it became clear early on that many other structures, especially those solved as part of drug-development efforts and neither fully refined nor published, might ultimately be lost. This prompted the National Institute of General Medical Sciences (NIGMS) to award in 1996 an interagency agreement to the National Cancer Institute (NCI), where some of the initial structural efforts on HIV PR have taken place, to create a repository that would contain as many structures as possible, and not necessarily only the published ones. This decision led to creation of the Internetbased HIV PR Database (HIVdb). After 6 years spent on creating and curating this database, the NCI effort is terminating on September 30, 2002. After that date, the National Institute of Standards and Technology (NIST), in collaboration with PDB, will take over and continue the project. This change of guard seems to be an opportune moment to remind the community about the existence of the database and the associated tools that have been created to enable its utilization. HIVdb is an Internet-based archive of experimentally determined three-dimensional structures of HIV-1, HIV-2, and SIV PRs and their complexes with inhibitors or products of substrate cleavage. HIVdb was one of the first databases of macromolecular structures created outside of the PDB. HIVdb includes both primary structural data and the derived information for this family of three very closely related enzymes. For that reason, it serves as an example of a special subset of a general structural database that can exist on its own, as well as coexist with a larger structural database and its tools. Information regarding one particular enzyme can show in detail how the structure adapts to binding of different ligands through changes in protein–ligand interactions, and conformationally adjusts to binding under different conditions. For proteins that serve as drug-design targets, it is important to study these interactions fully and in as many complexes as possible. Careful analysis of the wild-type as well as drug-resistant mutants of HIV PR may also help in creating new drugs that would overcome the problem of resistance. The structures contained in HIVdb have either been previously deposited in the PDB, or have been obtained directly from depositors for their inclusion in HIVdb only. In the latter case, they may have been less completely refined, or even not refined at all beyond the placement of the ligand; or they may have resulted from experiments that were never fully completed and published, but nevertheless were comparable in quality to the structures deposited in PDB. Many structures unique to HIVdb came