Structural Biology Symposium of the Sealy Center for Structural Biology Galveston, TX, March 19-21, 1999
The committee for the fourth annual symposium, chaired by James Lee and Bob Fox of the UTMB, organized yet another fascinating series of lectures in structural biology. This year, the major themes were an exploration of protein folding pathways (and where they go wrong!), signaling pathways that depend on protein-protein intreractions, and protein design. In the concluding session, plans to automate structure determination and modernize analysis were presented that should take us well into the next century.

Chris Dobson

Protein folding and misfolding
Chris Dobson (Oxford) opened the symposium with a discussion of protein aggregation and misfolding in disease. For years, aggregated proteins were not considered suitable for study by biophysicists, as most of the methods for structure determination require soluble protein or ordered crystals. This attitude changed as evidence accumulated that aggregated proteins are present in, and possibly the cause of, serious diseases, ranging from Alzheimers to Type II diabetes and Finnish hereditary amyloidosis. Mutations of lysozyme, used by Dobson's group as a model, are found in a hereditary amyloidotic disease. Wild type lysozyme, and even proteins not associated with disease, can be driven to make fibrils under the "correct" (low pH) solvent conditions. Fibril formation in these solution conditions is self propagating and can be accelerated by seeding with preformed fibrils, a fact that has previously been suggested as a mechanism for prion infectivity. In this view, fibril formation in disease results from a failure of cellular controls that should prevent proteins from accumulating in cellular compartments where they can unfold. Amyloid fibril organization in EM-photos (from Helen Saibil's group) would be consistent with, possibly intermolecular, ß-strand packing. Dobson concluded that protein fibrils reorganize after unfolding to tight ß -packed aggregates regardless of the initial native structure. For example, insulin, a predominantly a-helical protein, unwinds to form fibrils that, judging from NMR and FT-IR data, are primarily b-sheet (or b -ladder).

However, some proteins retain native conformation even when completely denatured by standard measurements. For example, apomyoglobin forms discrete folding intermediates at low pH, and Peter Wright's group (Scripps Institute) has shown with NMR measurements that these contain areas of native secondary structure, even at pH 2.

Their work represents an elegant solution to the major problem in studying protein folding pathways, i.e., isolating and determining the structure of intermediates when a protein folds in a few milliseconds.

Although the HSQC spectra of the intermediates are complex, the N and C resonances of labeled proteins are still well resolved, so their part of the spectrum can be used as reference. The average deviations of carbon chemical shifts from those expected for a random coil were used as a direct measure of secondary structure. These deviations showed that 3 of the 8 helices are partly folded, even at pH 2, where the molecule has a radius of gyration (Rg) of 30 Å. By pH4, when the Rg has compacted to 23 Å, only the middle three helices show any disorder. At higher pH, the protein appears fully folded (Rg =17.5 Å), but there is still disorder in the F helix until the chromophore is added.

George Rose (Johns Hopkins) suggested that folding specificity can be distinguished from stability, i.e., "why does lysozyme not fold like RNase?" is a different question than "why does lysozyme remain folded in solution?" He proposes, on the basis of simple steric considerations, that protein conformation is more limited than previously realized. This idea was tested in simple simulations of protein folding, which suggest the unfolded protein is preorganized. That is, a large number of local interactions interspersed throughout the molecule constrain subsequent folding stages to a high degree. This "biased sampling" means that initial folding will be quite rapid, and the final configuration will be limited by clashes with neighboring residues. Monte Carlo simulations of peptides were used to uncover the context-dependent conformational propensities of residues in a number of test molecules. As a protein loses a great deal of conformational entropy on achieving the native state, energy functions that do not include terms for this can capture these contributions to specificity.

Axel Brunger (Howard Hughes Medical Institute and Yale University) introduced the fourth dimension of folding with high resolution crystal structures of complexes of the SNARE (SNAP receptor) proteins that mediate fusion of synaptic vesicles with membranes. The core of the synaptic fusion complex1, characterized by protease digestion, contains a parallel four helix bundle with areas from 3 proteins (Syntaxin, synaptobrevin, Snap 25) twisted so tightly together that they appear to be one protein. Within a leucine zipper-like domain at the center, the Arg 56 side chain of synaptobrevin binds to three glutamine residues located within 2.5 Å. As these residues are for the most part conserved throughout the SNARE family (now termed the "R and Q snares"), they may be crucial to the assembly and disassembly of the complex. Further, similar 4 helical bundle structures can be seen in many proteins involved in viral fusion, including low pH treated influenza HA2, MuLV Mo-55, Ebola GP2, and SIV and HIV gp 41. The conservation of areas involved in the quaternary structure formation means that SNARE complex formation is promiscuous. Specificity may come from interactions with small G-proteins involved in synaptic vesicle exocytosis such as rab 3a (rabphilin). The hexameric NSF (N-ethylmaleimide sensitive factor) can act as a "chaperone" to disassemble stable SNARE complexes.

Click for Next Page