MAD data collection at SSRL

1. Preparing the experiment
   1. Crystals with an anomalous scatterer
   2. Absorption edge and scattering factor tables
   3. Alternative sample to measure fluorescence
2. Starting the experiment
   1. Fluorescence scan
   2. Excitation scan
   3. Wavelength selection
   4. Distance and exposure time at different wavelengths
   5. Order of collection
   6. Wedge data collection
3. Data collection
   1. Minimizing systematic errors
   2. Completeness of anomalous pairs
   3. Inverse beam
4. Appendix: Emission lines

Introduction

The MAD (Multiwavelength Anomalous Dispersion) phasing method exploits the abrupt changes in scattering power of heavy atoms in the vicinity of absorption edges. The changes in scattering power result in differences between the diffracted intensities measured at different wavelengths as well as differences between reflections of the type (hkl) and (-h-k-l) (anomalous pairs) measured at the same wavelength. This document explains how different experimental approaches affect the accurate measurement of these differences and provides hints to select the most effective data collection setup in most common experiments.

1. Preparing the experiment

1.1. Crystals with an anomalous scatterer

Elements with atomic numbers between 26 and 38 and between 62 and 92 are suitable for optimized MAD or SAD experiments ("optimized" means that the data are collected at the wavelengths where the anomalous effects are largest) and any atom heavier than sulfur can potentially give enough anomalous signal to solve the structure by non-optimized SAD.

1.2. Absorption edge and scattering factor tables.

They can be helpful to decide and double-check the choice of wavelength, particularly in non-optimized (off-edge) SAD experiments, L edge experiments or multiple anomalous scatterer experiments (SeMET-metalloproteins, TaBr clusters, etc). See for example http://www.bmsc.washington.edu/

1.3. Alternative sample to measure fluorescence

In practically all the cases, it is possible to obtain an adequate measurement of the absorption edge measuring the fluorescence from the crystal. In very special circumstances (for example, very tiny crystals with small anomalous/elastic scattering ratio, very radiation sensitive or unfrozen crystals) it can be practical to prepare an alternative sample for the fluorescence scan . A bunch of bad crystals mounted together is best for this purpose. Protein in solution tends not to be so good (the sample concentration in crystals is most often better). Heavy atoms in concentrated solutions can have different chemical environments than when bound to protein which can cause a shift in the absorption edge position. For this reason, it is convenient to "wash" derivative crystals in a mother liquor solution, so that only the signal from protein bound heavy atoms is measured.

2. Starting the experiment

2.1. Fluorescence scan

In most cases, this is the first thing you will do for MAD and optimized SAD experiments. To do the scan, go to the "Scan tab" in the program BLU-ICE and select the desired edge on the Periodic Table For the lighter anomalous scatterers (Iron to Strontium) you will select the K edge to perform the scan. For heavier elements it is recommended to select the L3 edge, because it is the one with the largest absorption jump (minimum f'). In the case of lanthanides such as Samarium or Europium, it is convenient to perform a scan over the L2 edge too. Although the L2 edge is smaller than the L3, the f" value is very large too and the shorter wavelength is better in terms of intensity and absorption errors.

Once you have selected your scan edge, click on "MAD Scan". This option will perform a wide scan about the absorption edge, then will calculate the scattering factor f" and f' for the scan range and output three wavelengths: The maximum f" and f' wavelengths in the scan range and a remote wavelength.

2.2 Excitation scan

If you want to make sure that your sample contains a heavy atom, or are searching for for metals centers or certain ligands in your protein you can do an excitation scan from the BLU-ICE "Scan tab". The difference between a "MAD scan" and the "excitation scan" is that the MAD scan measures the fluorescence counts from a single element in the sample while changing the energy and the excitation scan measures the fluorescence counts from different elements without changing the energy. The peaks in the excitation scan are known as "emission lines". These emission lines are characteristic for each element, therefore you can use the theoretical values for the fluorescence energy given in the appendix to id the elements in your sample. If your element of interest in not included in the appendix, search in the complete emission line tables

Remember that, as in the case of the fluorescence scans, detecting a signal from an element does not mean that it is bound to the protein. On the other hand, if you find a mysterious piece of density in your maps, the information from the excitation scan can be useful to solve the riddle.

2.3. Wavelength selection

The program BLU-ICE will automatically select the wavelength calculated from the fluorescence scan as the default ones for data collection as soon as you create a new run and click on "update". You should only override the selection in some special circumstances, for example:

1. You have decided to collect the maximum f" and minimum f' at different edges (in lanthanides).
2. You have several anomalous scatterers in the sample which affect the program estimate of f' and f" values.
3. In the particular case of Iron, it can be advantageous to select a remote at somewhat higher energy than recommended by the software, to take advantage of the higher beamline intensity and higher resolution achievable at higher energies. A remote energy about 10000 eV works well.

2.4. Distance and exposure time at different wavelengths

The optimal value for these two parameters will change with the wavelength. Optimizing the distance near the absorption edge and using the same value for the remote wavelength means that the whole area of the detector will not be used efficiently at this wavelength (the diffraction pattern will be compressed towards the center of the detector. This could affect dramatically the quality of the data, particularly if the MAD experiment is on a L edge element (because the edge and remote wavelengths will be very far apart) or if the crystal has a very large cell (because the number of overlaps will increase at the shorter wavelength). Use the resolution widget in the BLU-ICE hutch tab or the program "xtalcalc" to calculate the correct distance for each wavelength. Note: This does not mean that you should collect to the same resolution limit at all the wavelengths: If the crystal diffracts to high resolution it may be a good idea to collect one of the wavelength to higher resolution to be able to extend the phases and obtain more interpretable maps, but you should make good use of the detector active area.

The beamline intensity also changes at different energies. However, because the background noise also decreases with the intensity, the quality of the data is likely to be similar at all energies, except for the lowest ones (near the absorption edges of Iron, Nickel, etc). In this case, particularly if the crystals are weakly diffraction, it is convenient to collect one image at one of the near edge wavelengths (either the peak or inflection) and at the remote using the same dose and compare the diffracted intensity/background at the two wavelengths. If it is very different, increase the exposure time for the "weaker" wavelength.

Currently BLU-ICE does not support collecting at multiple detector distances or exposure times within the same run. If you want to do this, you have to create a separate run for each wavelength. The easiest way to do this is as follows:

1. Create a new run and click "update" to select all the wavelengths. Edit the run: Enter the root image name, phi and delta phi and the distance and time for the first wavelength. Do not remove the other two wavelengths!
2. Create two additional runs. This will clone the run you have just prepared twice.
3. For each run, delete the appropriate wavelengths and enter the correct time and distance.

2.5. Order of collection

The default order of collection (maximum f", remote, minimum f') is devised to provide the maximum amount of phasing information as the data are being collected, so that it is possible to obtain good quality phases if the data set cannot be completed for any reason. If the remote wavelength has a higher f" than the edge peak it is better to collect at the remote wavelength first, followed by the edge inflection point (minimum f'). This can be the case for some L edge experiments such as Platinum, Mercury, Gold or Lead. The edge peak wavelength contribution to the phasing is not so important in this case and it is better to skip it unless the signal is very poor.

If the fractional anomalous signal in your sample is high (for example, if there are 3 or more non N-terminal Selenium atoms per 100 residues) it can also be advantageous to collect the remote wavelength first, followed by the inflection wavelength. In this case, it is possible to measure a large anomalous signal at the remote wavelength. Also, data collection at the remote wavelength results in a lower radiation dose absorbed by the crystal and slower onset of radiation damage.

2.6. Wedge data collection

BLU-ICE offers the option of automated MAD data collection by small angular wedges, rather than collecting the complete wavelengths. one by one. Data collection in this mode is useful to collect good dispersive differences when the crystal experiments a fast decay due to radiation damage. This data collection mode is indicated for small poorly diffracting crystals, which always receive a relative larger dose than crystal requiring shorter exposures. In these cases it can be convenient to do the wedge data collection over only two wavelengths (remote and minimum f'). These two wavelengths combination gives very good MAD phases (practically equivalent to three wavelength ) because of the optimized dispersive difference. The maximum f" wavelength can be collected last if the crystal is still diffracting.

3. Data collection

Most requirements for a high quality data collection (completeness, avoiding overloads and overlapped reflection, using the whole detector surface, etc) apply to MAD or SAD data collection as well. In addition, there are some particular requirements to collect good accurate anomalous and dispersive differences:

3.1. Minimizing systematic errors

Being able to collect data with the crystal at the same orientation for all the wavelengths minimize systematic errors on the differences and it is very important for the success of MAD experiments. If you think there is a chance that the crystal orientation will change or experimental conditions will change during data collection (your crystal is in a capillary and seems to be slippage, there is some ice build-up around the crystal and you will have to touch the pin or loop to clean it, etc) you should collect "wedges" at at least two different wavelengths (inflection and remote or maximum f" and remote) so that the differences between these wavelengths are not affected. Selecting wedge mode can affect the data resolution at one of the wavelengths, because of the different optimal sample-detector distance at different wavelengths

3.2. Completeness of anomalous pairs

Because only half of symmetry related reflections are anomalous (or Bijvoet) mates of a given reflection, the rotation range which gives the most complete Bijvoet set may not be the same which gives the most complete data set. You can find out what the optimal angular range is by running the strategy option in mosflm, but instead of typing

strategy go

you need the command:

strategy anomalous go

Note!: If you do not have enough time to do finish a MAD experiment, it is better to sacrifice some anomalous completeness and try to collect full data sets at at least two wavelengths: It is often still possible to do successful MAD experiments with 70-80% anomalous completeness, but only rarely with less than 90-95% overall data completeness.

3.3. Inverse beam

An alternative way to collect a complete Bijvoet difference set is to use the range given by the normal mosflm strategy option and collect data over than range and 180 degrees away (reflections related by inversion are Bijvoet mates). BLU-ICE has an option to collect inverse beam data. You have to enter the direct beam phi range and the program will generate the input to collect the inverse beam range automatically.

Beware!: Using the inverse beam option in BLU-ICE can be disadvantageous for low symmetry space groups, as the crystal may suffer from radiation damage at the first wavelength. As a precaution, it is better to collect half of the data at all the wavelengths and then collect the inverse beam pass on a different run. Using dose mode will ensure that the "inverse" pass will have the same exposure.

Inverse beam is recommendable when the rotation range required for a complete anomalous pairs data set is twice the one necessary for a non-anomalous data collection. This is the case for triclinic and trigonal space groups and for certain crystals orientations in monoclinic, tetragonal (point symmetry 4) and hexagonal (point symmetry 6) space groups.

For orthorhombic and 422 and 622 point symmetry space groups, where the same total range is required as for non-anomalous data collection it is preferable not to use this option, unless the expected signal is low or the anomalous Patterson is impossible to interpret. In these situation, increased multiplicity can provide clearer Patterson maps.

Beware! The inverse wedge will not significantly improve the completeness of your data, but good completeness is critical to MAD experiments! If your total data completeness completeness hovers around 90% , collecting 10 extra degrees can give better results than collecting inverse mode.

Appendix: Emission lines

Strongest emission lines for the most common elements present or introduced in biological macromolecules. (Emission and excitation energies in eV).