Two-wavelength MAD phasing: in search of the optimal choice of wavelength

Gonzalez et al. [Acta Cryst. D55 (1999), 1449-1458] reported in their recent article that interpretable electron-density maps, similar in quality to those calculated with data collected at three wavelengths, could be obtained using only two wavelengths if the wavelengths were chosen so as to give a large contrast in the dispersive component of the scattering factor. Four different crystals, which contain iron, gold, iridium, or selenium atoms, were used in their study.

The multiwavelength anomalous dispersion (MAD) method exploits the structure-factor variation with wavelengths around the absorption edges of heavy atoms within the protein crystals. The variation consists of the real part (or the dispersive component, f') and the imaginary part (or the anomalous component,  f''). The importance of using wavelengths that can provide the largest f' was explained as follows: “Firstly, the refinement of the anomalous scatterer positions and occupancies is highly dependent on the dispersive differences measured from centric reflections. Secondly, a MAD data collection is usually performed in a way which partially cancels out systematic errors in the structure factors at different wavelengths, leading to dispersive differences which are less affected by errors in the data.”

Two approaches have been used in calculating phase information from a MAD experiment. One approach uses an analytical method to solve the phase problem and calls for three or more wavelengths to optimize the results. The other approach treats MAD as a traditional isomorphous replacement phasing. The importance of selecting a large difference in the real part of the scattering factor in the two-wavelength experiment is interesting, as it resembles what is known from the single isomorphous replacement and anomalous scattering (SIRAS) experiment, i.e., phasing power increases with increased differences in the real part of the structure factors, which is achievable by the incorporation of heavier heavy atoms.

Most structures determined by the MAD method thus farwere obtained from data collected at three or more wavelengths. Doing the MAD experiment using fewer data sets will reduce crystal exposure to X-rays and is a good option for structure determination using a high-flux synchrotron source or as an onbeamline map calculation whereby a two-wavelength map is calculated to evaluate the need, on the fly, for collecting data at more wavelengths. More interesting still, is the one-wavelength approach for on-the-fly map calculation using data from the veryfirst wavelength, where radiation damage is the least. Using a numerical method for resolving phase ambiguity, we have recently determined the structure of an Fe-containing protein (86kDa per asymmetric unit) using single-wavelength iron anomalous data collected in-house (P12.02.023 IUCr 99 program abstracts). For single-wavelength anomalous scattering data, collection at the absorption edge is preferable. However, the use of other wavelengths, in-house or at synchrotron, is an option provided that the measurements are accurately made.

Bi-Cheng Wang
U. of Georgia, Athens, GA, USA