Research

The Wedemeyer lab combines computational and experimental methods to answer important biological questions about protein structure. Our principal focus is gp120 (the envelope protein of HIV) but we are also pursuing basic research about the physics of protein folding using peptides and small proteins. Our software is protected by the GNU Public License.

Conformational Changes in gp120 During HIV-1 Cell Entry
HIV is perhaps the most critical public health problem of our time, with roughly 40 million people infected and roughly 10,000 deaths per day. The HIV envelope protein, gp120, is the principal target for vaccines and antibody neutralization, being the sole exposed viral protein. However, gp120-based vaccines have not been successful and relatively few anti-gp120 antibodies are effective at neutralizing the virus; of these few antibodies, none are effective against all forms of the virus.

We hypothesize that gp120 avoids neutralization by cloaking the structures needed for infection until the moment of infection. gp120 undergoes at least two major conformational changes during viral entry, the first upon binding to CD4 (its receptor on the target cell) and the second upon binding to a cellular chemokine receptor (either CCR5 or CXCR4). These conformational changes in gp120 may produce the structures needed for infection, either (1) by uncovering pre-formed but buried structures or (2) by folding those structures from a disordered or differently folded conformation. Unfortunately, these conformational changes are poorly understood, since gp120 is difficult to crystallize and too large (approx. 477 residues) for traditional NMR.

We are tackling this problem by combining de novo protein-structure prediction methods with medium-resolution structural probes, including 19F NMR, fluorescence, chemical modification, and proteolysis. Our gp120 comes from a primary isolate of clade A2 HIV-1, and has been cloned and expressed in mammalian and insect expression systems. Virological experiments are being carried out in parallel by our collaborator, Mary Poss.

Computational Protein Structure Prediction/Design
The methods of computational protein structure prediction have matured in recent years, so that the structure of individual domains can be predicted de novo (i.e., from the amino-acid sequence alone) within 3-5 Å CA rmsd. Nevertheless, these methods are prone to fail if the domain is unusually large (>150 residues), has poorly predicted secondary structure, or has a high fraction of long-range contacts. Moreover, there are no reliable methods for refining approximately correct structures (e.g., 5 Å CA rmsd to native) to higher resolution (e.g., 3 Å CA rmsd). We are developing new sampling methods and force-fields to overcome these obstacles.

Experimental Studies of Protein/Peptide Electrostatics
Successful protein prediction and design requires accurate energy functions for scoring trial conformations. Unfortunately, the electrostatic component of present-day energy functions seems to be inaccurate; this is especially problematic, since the electrostatic component is long-ranged. We are studying the electrostatics of peptides and proteins experimentally in an effort to improve these potential functions.

Publications

Search PubMed at the National Library of Medicine: