The “Specimen Flatness” Problem in High-Resolution Electron Crystallography of Biological Macromolecules

Abstract

Methods of image acquisition and image processing have been improved to such an extent in recent years that it is now possible to obtain electron micrographs of well ordered monolayer crystals of biological macromolecules at 3.5 Å resolution or better. Similar images must be recorded at high tilt angles, preferably up to 60°, in order to reconstruct a three-dimensional density map with comparable resolution. In order to record images at the highest tilt angles it is essential that the specimen be almost perfectly planar, perhaps to a tolerance as low as ±0.1°. In the case of crystalline arrays of bacteriorhodopsin, such a high degree of specimen flatness, or planarity, is seldom achieved by current techniques of specimen preparation, and this fact is a major impediment in advancing the current status of the structure analysis. It is likely that similar problems in achieving adequate specimen flatness will be encountered with monolayer crystals of other biological macromolecules.A physical analysis of the role of surface tension forces and interfacial energies in the preparation of glucose-embedded specimens suggests that flat specimens could hypothetical be obtained on either hydrophilic or hydrophobic carbon films. In practise, we have so far had success with hydrophilic carbon in reliably obtaining strongly diffracting specimens of purple membrane that are flat to better than ±0.5° only when a glow discharge in H2O vapor was used as the process for creating a hydrophilic surface (Figure 1). While this represents progress, the result is still not as good as it needs to be. When using hydrophobic carbon, we have obtained a striking improvement in the consistency of our results by applying the specimen to the wet side of the grid, immediately after picking up a small square of carbon from the surface of a glucose solution with a 400 mesh grid. After mixing, some of the excess solution and sample is removed with the pipette, and the remainder is blotted off by pressing the grid bars directly onto filter paper (see Figure 2). Freshly evaporated carbon can be used immediately after converting it to a state of increased hydrophobicity by heating the carbon, on mica, in a drying oven at 80°C or higher, for a period of one hour or more. Some optimization of the glucose concentration is needed, ranging from 2 percent to 20 percent, apparently depending on residual variation in the surface properties of the carbon film. In the best case up to half of the membranes that “look” flat are indeed flat enough to give diffraction patterns that are isotropically sharp at tilt angles up to 60° (Figure 3). In virtually all cases, however, an acceptable fraction, rarely less than one percent, of the membranes that look flat are, in fact, extremely flat. A new problem encountered at this point, however, is that the specimens become substantially wrinkled when cooled to -120°, as evidenced by the fact that the diffraction pattern is no longer isotropically sharp at high tilt angles. Current research is investigating whether this latest problem can be overcome by using hydrophobic polymer films, rather than carbon, as their coefficient of thermal expansion should be more similar to that of protein crystal than is the case for evaporated carbon.