Perspectives for the Future

The impressive expansion of the computational power of computers is at the core of advances in determining the three-dimensional structures of proteins. Vacuum-tube computers running on programs punched on cards were used to calculate the first protein structures from x-ray crystallographic images, a process that at the time took years, but can now be accomplished in a matter of days and in some cases hours. In the future, researchers hope to predict the structures of proteins using only amino acid sequences deduced from gene sequences. This computationally challenging problem requires supercomputers or large clusters of computers working in synchrony. Currently, only the structures of very small domains containing a hundred residues or fewer can be predicted at a low resolution. However, continued developments in computing and models of protein folding, combined with large-scale efforts to solve the structures of all protein structural motifs by x-ray crystallography, will allow prediction of the structures of larger proteins. With an exponentially expanding database of structurally defined motifs, domains, and proteins, scientists will be able to identify the motifs in an unknown protein, match the motifs to the sequence, and use this information to predict the three-dimensional structure of the entire protein.

New combined approaches will also help in determining the structures of molecular machines at high resolution. Although these very large macromolecular assemblies are usually difficult to crystallize, and it is therefore difficult to determine their structures by x-ray crystallography, they can be imaged in a cryoelectron microscope at liquid helium temperatures and high electron energies. From millions of individual “particles,” each representing a random view of the protein complex, the three-dimensional structure can be built. New methods have recently dramatically improved the resolution attainable using cryoelectron microscopy, particularly when used to study membrane proteins. In some cases, the structures of individual subunits of a large multiprotein complex may already have been solved by x-ray crystallography, permitting a composite structure consisting of the x-ray-derived subunit structures fit to the cryoelectron microscopy-derived model to be generated.

Methods for rapid structure determination, combined with identification of novel substrates and inhibitors, will help determine the structures of enzyme-substrate complexes and transition states and will thus help provide detailed information regarding the mechanisms of enzyme catalysis. Membrane proteins, because of the specialized environment in which they reside and their solubility characteristics, remain challenging, although progress in this area is accelerating.

Although our understanding of chaperone structure and activity continues to grow exponentially, a number of critical questions remain. We do not understand precisely how cells make the distinction between unfolded and misfolded versus properly folded proteins. Clearly the exposure of hydrophobic side chains plays a role, but what are the other determinants of this key recognition process? How is the decision made to turn from trying to refold a protein to degrading it?

The rapid development of new technologies can be expected to help solve some of the still outstanding problems in proteomics. It is becoming possible to identify and characterize intact proteins as large as 30–70 kDa in complex mixtures using MS techniques without first digesting the samples into peptides—a method called the “top-down” approach, in contrast to starting with fragments of the protein (“bottom-up” approach). An ongoing problem in proteomic analysis of complex mixtures is that it is difficult to detect and identify protein fragments whose concentrations in a sample differ by more than 1000-fold: some samples, such as blood plasma, contain proteins whose concentrations vary over a 1011-fold range. Routine analysis of specimens with such diverse concentrations should dramatically improve the mechanistic and diagnostic value of blood plasma proteomics. New specimen handling methods will permit comprehensive proteomic analysis of much smaller samples (5000 cells or less). Improved sensitivity will open up to proteomic analysis the investigation of rare types of cells and small samples of normal and diseased tissue.