W0437

Trypanosomal pyruvate phosphate dikinase (PPDK) is a potential target for the development of new drugs against several tropical diseases. Purification of PPDK¹ yielded only one batch from which crystals grew. Myriads of tiny crystals appeared in seconds with pH close to the pI of PPDK. Improvement was done by adding minute amounts of PEG, and by the use of glycerol as crystal growth retardant². We collected at LURE a complete 3.0 Å resolution data set at 4^o using all available crystals. Hence, only molecular replacement (MR) was feasible. The sole model available was bacterial PPDK³. The work done on this homo-dimeric enzyme had showed 2 remote domains per subunit. One domain contains the site for ATP/AMP binding while the other binds pyruvate/PEP. A third small domain shuttles a His-bound phosphate between the 2 active sites during catalysis. Domain motion is thus part of the mechanism of PPDK.

Attempts to use the full subunit for MR were unsuccessful because of the different relative orientations of the domains. We solved the MR problem piecemeal, placing domains and subdomains one at a time: the largest C-t domain (342 residues) was positioned (cumulative Rfree=0.492). The N-t domain was then placed (cRfree=0.471); the resulting map had clear density for its C-t segment (127 residues) but no density for the N-t segment (240 residues). Hence, these two segments were placed separately, first the Cts (cRfree=0.459) and then the Nts (cRfree=0.432). Attempts to use "classical MR" to locate the phospho-Histidine domain (12% of the electrons) failed, although the maps showed uninterpretable positive density for this missing fragment. Hence, we used "real space" MR⁴ to locate the last domain: estimated structure factors for the missing fragment were generated by a density modification procedure called "protein flattening". Here, the coordinates for the partial model were expanded to cover one unit cell. From these, an envelope was generated, and used to flatten the density corresponding to this partial model. What remains thus corresponds to the missing fragment, and to the solvent. Map inversion provided the data for rotation- and phased translation- functions. Application of the procedure led to the positioning of the phospho-His domain (2^nd peak in the list, cRfree=0.384) and hence to the 3D structure.