Page 60

There are two complementary approaches to the characterization of unknown proteins in structural genomics projects. First is the “low hanging fruit” approach, in which proteins are cloned, expressed, purified and structures solved, but, initially, at each step only the proteins that behave well and are easy to work with are carried forward to the next step. Second is the rational preselection of representatives of predicted new fold classes, or proteins that may have unusual physical properties, and a concentrated focus on characterizing these proteins, regardless of how easy or difficult it is to work with each protein. Each approach has a number of advantages and disadvantages, and at this stage in structural genomics research, there is clearly a need for both.

As an example, let's consider our work on MG (Balasubramanian et al., 2000). MG has the smallest genome known for any self-replicating organism, encoding approximately 483 proteins (the exact number of predicted proteins varies slightly, depending on the identification method used). Of these, 202 are structurally uncharacterized, 70 are both functionally and structurally uncharacterized, and 25 are completely structurally and functionally uncharacterized over their entire length. Of these 25, 15 are unique to MG, and 10 have homologues in related organisms. We have expressed, purified, and characterized 12 of these 25 proteins. Seven behave like “normal” proteins, display substantial secondary structure, and likely represent novel folds. These are candidates for high-resolution structure determination. One protein is unstructured and may require a partner molecule (either another protein subunit or a nucleic acid or other cellular component) in order to fold, and two display unusual thermodynamic properties: they are highly helical and extremely resistant to thermal denaturation. These latter two proteins are highly conserved from MG to man.

What stages in this work provide engineering challenges and opportunities? Currently, the most common means of production of the large amounts of protein required for biophysical and structural characterization is expression in live bacteria. A better high-throughput method would speed the process. Purification of large quantities of protein to a reasonable level of purity can be accomplished quite readily by attaching a universal “tag,” which allows all proteins to be purified by that same method, regardless of their individual chemical nature. Development of an additional universally applicable purification step, which would easily allow proteins to be purified to the high levels required for crystallization, is important. At the moment, a more significant problem than these is that at least half of all proteins, when expressed in bacteria, do not partition into the soluble phase but, rather, aggregate and partition as unfolded protein into the insoluble fraction. This means that they must be refolded before characterization can be begun, or bacterial growth conditions must be manipulated to “coax” as much protein as possible into the soluble phase. In our work with the uncharacterized proteins of MG, obtaining reasonable quantities of pure folded protein certainly was the rate-limiting step.