Electron diffraction analysis of the M412 intermediate of bacteriorhodopsin.

High resolution electron diffraction data have been recorded for glucose-embedded purple membrane specimens in which bacteriorhodopsin (bR) has been trapped by cooling slowly to below--100 degrees C under continuous illumination. Thin films (OD approximately 0.7) of glucose-embedded membranes, prepared as a control, showed virtually 100% conversion to the M state, and stacks of such thin film specimens gave very similar x-ray diffraction patterns in the bR568 and the M412 state in most experiments. To be certain that any measured differences in diffraction intensity would be real, two independent sets of electron diffraction intensities were recorded for near-equatorial, i.e. (hkO), reflections. Little correlation was indeed observed between these two sets for delta F values at low resolution (15-5.0 A, 49 reflections), but the correlation coefficient is approximately 0.3 at high resolution (5.0-3.3 A, 218 reflections). Thus, while most of the measured difference is error, the mean delta F and the correlation coefficient can be used to estimate the smaller, true delta F due to structural changes occurring in the M state. The magnitude of this estimated true mean delta F is equal to what would be produced if approximately five to seven nonhydrogen atoms were moved to structurally uncorrelated (i.e., new) positions in the M state. Movements of a few amino acid side chains, and repositioning of atoms of the retinal group and the associated lysine side chain after trans-cis isomerization, are the most probable causes of the observed intensity changes in the M state. The difference Fourier map, calculated in projection at 3.5-A resolution, shows only very small peaks, the largest of which are confined, however, to the region of the protein.