Drug Development

Structure-guided drug discovery

The lab is both working on the development of tools to facilitate the use of protein structure information in drug discovery, and the application of such tools to several actual targets that are important form human diseases.

Tool development and applications:

1) In silico high-throughput screening

In silico screening

The very high cost of high-throughput screening (HTS) for small molecules that bind to a given macromolecular target, a time and labor-intensive process, has generally prohibited academic or non-profit groups from serious drug discovery programs. Advances in structural biology and computer-based modeling have recently provided a set of new tools to probe the interaction between proteins and various molecules (including other proteins or potential small molecule inhibitors). Thus, it is now feasible to undertake high throughput screens (HTS) for drug candidates primarily in silico by docking a library of compound structures to the appropriate site(s) on the target protein in the computer. This can greatly enrich the pool of candidate molecules for actual experimental screening. For example, whereas a typical HTS has a hit rate of less than 0.05%, HTS using a compound collection that has been prescreened in silico to contain only those compounds predicted to fit into the desired site and interact with the target with reasonable affinity will often have a hit rate of 10-15%, nearly a thousand-fold enrichment. This means that instead of having to screen 100,000 compounds to obtain 10 confirmable hits, an academic lab need only screen ~100 – a manageable workload even with complex, cell-based assays.

An in silico HTS is a computer simulation of an actual experimental drug screen in which a very large library of candidate compounds can be "screened" by docking models of their structures into the active site of the target and then calculating the energy of interaction between the compound and the protein to identify those molecules that show a good fit for the target site.
Whereas an actual HTS directly selects on the basis of the presence of a biological activity - for example, enzyme inhibition - an in silico method merely screens for compounds most likely to bind; hence, in silico screening requires subsequent in vitro and in vivo testing of the identified compounds. (Kitchen et al., 2004 and references therein).

The steps involved in in silico modeling are the following:

  1. Generating a structural model of the target macromolecule, either by means of homology modeling from existing, similar experimentally-determined structures or by actually determining the structure directly. Hydrogen atoms must be added to the model computationally as these are not normally observed in an X-ray diffraction experiment. In the case of several of the protein components of the retromer complex, excellent crystal structures are available.

  2. Identifying the possible binding sites. In the case of inhibitors, this will usually (though not always) be the known active site (though it could be an allosteric inactivation site). For activators, allosteric sites are preferable. If the protein is under allosteric control, then such sites are also known. If the protein is not, then it is necessary to find sites where the binding of compounds will stabilize the structure, thereby increasing the protein's stability in vitro and in vivo. Increasing the half-life of the protein is equivalent to overexpression or direct chemical activation, since more of the active protein exists in the cell at steady-state (see Figure 1 for an example). To identify all possible binding sites on the surface of a target protein, we have developed two new methods: MSCS (also known as "solvent mapping") and THEMATICS. THEMATICS is a purely computational approach that is applied to the atomic coordinates of the target protein (Ondrechen et al., 2001). MSCS may be carried out either experimentally or computationally (Mattos et al., 2006). Both have a very high success rate.