Electron crystallography of soluble proteins

Protein structures at moderate to high resolution can now be readily solved using an electron microscope and image processing techniques. The averaging conditions required to overcome the low signal to noise ratio of electron microscope images are particularly well satisfied by two-dimensional crystals. However, a major limiting step in such a study is production of suitable protein arrays.

The work presented describes a study of the two-dimensional crystallization and structural characterisation of several biologically significant soluble proteins. An interfacial crystallization technique was employed where proteins congregate at the interface between an aqueous subphase and a phospholipid layer due to interaction with the phospholipids. The inherent fluidity of the layer allows orientation and aggregation into a two-dimensional lattice.

A Langmuir trough investigation of the surface activity of the 33 kDa Staphylococcus aureus pore forming [alpha]-toxin and its interaction with phospholipid layers at the air/water interface showed the protein to have a detergent like behaviour and a strong affinity for the negatively charged phospholipid dimyristoylphosphatidylserine. Optimising the parameters of this interaction led to successful twodimensional crystallization of the toxin into a tetragonal lattice with two-sided plane group symmetry p4212. Tilt series data collected from such crystalline areas were used to reconstruct a three-dimensional structure of the toxin oligomer at ~17 Å resolution (Paper I).

A similar interfacial approach to the two-dimensional crystallization of native Chaperonin TF55 from Sulfolobus solfataricus, hyperthermophilic archaea, yielded two crystal forms of this large (1 MDa) hetero-oligomeric double ring complex. A phospholipid dependent orientation of the oligomer was observed. Side-view para-crystalline areas arose from interaction with dimyristoylphosphatidylcholine whereas interaction with dioleoylphosphatidylglycerol produced trigonal arrays of the complex in an endon orientation.

The two-dimensional projection structure calculated from the trigonal arrays exhibited the symmetries of two-sided plane group p312 allowing us to suggest a three-dimensional arrangement of the two subunit types (Paper II). This model was later confirmed and enlarged upon via tilt-series data collection, and reconstruction into a three-dimensional volume at ~18 Å resolution (Paper IV). A projection map from the fortuitous two-dimensional crystallization of reconstituted [beta]-rings of the chaperonin TF55 from the closely related Sulfolobus shibatae also showed p312 symmetry even though the [beta] subunit oligomers possessed, as expected, a non-crystallographic nine-fold symmetry axis (Paper III).

Application of these Langmuir trough interfacial techniques to the two-dimensional crystallization of two steroid-hormone receptor constructs (Estrogen receptor (alpha) ligand binding domain and a Protein-A glucocorticoid receptor DNA binding domain fusion protein), resulted in successful crystallization of one of the proteins. The approach's relevance as a screening method for two-dimensional crystallization was demonstrated (Paper V).