Page 135 - Macromolecular Crystallography

124 MACROMOLECULAR CRYS TALLOGRAPHY

symmetry averaging of the density due to the two of MAD phases is the fact that the data come
copies in the asymmetric unit. A model of the com- crystals that approximate true isomorphism more
plete molecules could easily be built into a map closely than MIR data. Nevertheless, there might
computed with these phases (Murthy, 2001). be cases in which the anomalous signal might be
Althoughheavyatomoranomalousscattererposi- too weak to provide a complete solution, although
tions have traditionally determined from Patterson approximate phases can be obtained. Classically, it
maps, alternative techniques are available. Espe- has been possible to combine structural informa-
cially in cases where the number of anomalous tion from more than one source through combina-
scatterers is large, direct methods have been used. tion of suitably weighted phases. This is generally
In fact, in the determination of the low resolution accomplished by multiplication of the probabil-
structure of Clostridium acdi-urici ferredoxin through ity distribution function for the phase from each
MAD phasing (Murthy, 1988), the positions of the source (Hendrickson and Lattman, 1970). A prob-
iron-sulphur clusters was located using MULTAN ability density function representation for MAD
(Germain et al., 1970). Because the data measured phases has been derived (Pahler et al., 1990) and
from the crystals were limited to about 5 Å, no such functions have long been available for MIR
attempt could be made to obtain individual Fe and and model phases (Blundell and Johnson, 1976).
0
S atomic positions. A selected set of | F A (h)| val- Currently the most widely used improvement tech-
ues were input to the program for computation of E nique for initial phases from any source is that of
values and to obtain a very straightforward solu- statistical density modification (Cowtan and Main,
tion for the centroids of the two clusters in the 1998; Cowtan and Zhang, 1999). SOLVE/RESOLVE
asymmetricunitbyusingthephasesetwiththehigh- has algorithms that implement such techniques
est combined figure of merit. Although Patterson (Terwilliger, 2001, 2002), as do programs in the CCP4
methods have been used to determine substruc- system (Project, 1994). Non-crystallographic sym-
tures containing as many as 50 atoms (Terwilliger, metry can also be used to improve the quality of
1997a), Direct Methods are becoming increasingly initial phases obtained from any source including
popular for this purpose. For example, the 15 MAD phases.
Se positions in selenomethionyl T7 DNA poly- MAD phasing has, over the past decade, become
merase were determined (Doublie et al., 1998) using one of the most widely used techniques to determine
SHELXS86 (Uson and Sheldrick, 1999). Newer pro- macromolecular structures. Its increasing appeal can
grams such as SnB (Howell, 2000) have also had be attributed to a succession of advances in the theo-
remarkable success in determining large substruc- retical foundations, data measurement, processing,
tures of anomalous scatterers in macromolecules. and analysis as well as parallel developments in
Thirty Se positions in selenomethionyl substituted refinement of initially obtained phases. Currently
S-adenosyl-homocysteine hydrolase (Turner et al., ongoing work, in extending data measurement at
1998), 48 Se positions in the selenomethionyl EphB2 synchrotrons to longer wavelengths, is likely to per-
receptor (Thanos et al., 1999), and 65 Se positions in mit the use of K absorption edges of native sulphur
selenomethionylADP-l-glycero-d-mannoheptose6- atoms in proteins and phosphorous in nucleic acids
epimerase (Deacon and Ealick, 1999) have all been as potential MAD signal generators.
successfully determined using SnB. The CCP4 pro-
gram system also offers direct methods based tech-
Acknowledgements
niques for determination of anomalous scatterer
positions through the RANTAN (Yao, 1981) and Thanks are due to Abdul Ajees for help in prepa-
ACORN (Yao, 2002) programs. ration of the manuscript and to Larry DeLucas for
In most cases, structures are determined from support and encouragement. The research in my lab-
MAD data measured on a single crystal. This makes oratory is supported by NIH grants AI45623 and
the processed data inherently more precise and accu- AI51615 and by a Focused Giving grant from the
rate, because scaling across crystals can be avoided. Johnson & Johnson department of Corporate Science
Asecond circumstance that contributes to the quality and Technology.