Diffraction data analysis in the presence of radiation damage

Abstract

In macromolecular crystallography, the acquisition of a complete set of diffraction intensities typically involves a high cumulative dose of X-ray radiation. In the process of data acquisition, the irradiated crystal lattice undergoes a broad range of chemical and physical changes. These result in the gradual decay of diffraction intensities, accompanied by changes in the macroscopic organization of crystal lattice order and by localized changes in electron density that, owing to complex radiation chemistry, are specific for a particular macromolecule. The decay of diffraction intensities is a well defined physical process that is fully correctable during scaling and merging analysis and therefore, while limiting the amount of diffraction, it has no other impact on phasing procedures. Specific chemical changes, which are variable even between different crystal forms of the same macromolecule, are more difficult to predict, describe and correct in data. Appearing during the process of data collection, they result in gradual changes in structure factors and therefore have profound consequences in phasing procedures. Examples of various combinations of radiation-induced changes are presented and various considerations pertinent to the determination of the best strategies for handling diffraction data analysis in representative situations are discussed.

Figure 1

Linear increase of the scaling (relative) B factor for various diffraction data sets: (a) APC35880, (b) BPTI, (c) NaI-841, (d) p37n33, (e) Tp0655, (f) VAV. In cases where the crystal was smaller than the beam size (APC35880, BPTI, NaI-841 and p37n33, Tp0655) there is little fluctuation in B-factor behavior. VAV represents a case where the crystal was larger than the beam and in which three data sets were collected, with the third data set being acquired after moving the crystal to a new position.

Figure 2

Distribution of height of RDDEM peaks at Se-atom positions for APC35880. The height of the peaks was scaled by dividing the height of each peak by the height of the highest RDDEM peak at a Se atom.

Figure 3

Radiation-induced specific changes around ligand-binding sites in p37n33 (a) and TP0655 (b). RDDEM contour levels are expressed in root-mean-square units (σ). The green color corresponds to the +5σ level and red to the −5σ level. For clarity, water molecules were relabeled with respect to PDB entries 3e79 and 2v84. For 3e79, W1 = W1035, W2 = W1002, W3 = W1046, W4 = W1028 and W5 = W1090; for 2v84, W1 = W116.

Figure 4

Differences in radiation-induced specific changes at disulfide bridges of BPTI. Red surfaces represent the −5σ level and gray surfaces represent the +5σ level. Cys51 is damaged at a much faster rate than other cysteine residues.

Figure 5

Lattice destruction in an APC5871 crystal. The figure shows a dramatic increase in mosaicity at image 180, suggesting progressive lattice destruction.