S0203

X-RAY STRUCTURE DETERMINATION OF HUMAN CERULOPLASMIN AT 3.0 Å RESOLUTION. G L. Card, I. Zaitseva, V. N. Zaitsev, B. Bax, A. Ralph, P. F. Lindley, CCLRC, Daresbury Laboratory, Warrington, WA4 4AD, UK

Human Ceruloplasmin (hCP) is a copper containing glycoprotein with a molecular weight of approximately 132 kDa, corresponding to some 1046 amino acid residues and 7-8% by weight carbohydrate. The precise functions and chemistry of the protein have not been defined, but it has been associated with ferroxidase activity, amine oxidase activity, anti-oxidant activity and copper transport, and it may indeed be multi-functional. Sequence analysis of the protein indicates a domain structure involving internal triplication and an extraordinary homology with the A-type domains of blood clotting factor VIII.

Preparations of hCP exhibit heterogeneity, particularly with respect to the carbohydrate moieties, and the protein is very susceptible to proteolytic cleavage, aggregation and loss of copper. Numerous modifications have been made to the isolation and purification procedures to obtain reproducible crystal for X-ray analysis. Recently a S200/300 gel filtration stage has been introduced to remove aggregated and proteolytically cleaved fragments immediately prior to crystallisation. This has enabled the production of trigonal crystals, spacegroup P3221 with a = 213.92 Å and c = 85.63 Å.

Using the SRS synchrotron source at Daresbury Laboratory, data has been collected for the native crystals and two heavy atom derivatives. Difference Patterson syntheses for both derivatives were readily interpretable in terms of single site binding. Protein phases were calculated using the program MLPHARE, and lead to an electron density map at 3.4 Å resolution which clearly showed the molecular boundary. Improvement of this map using DM, incorporating solvent flattening and histogram matching, led to an electron density map at 3.0 Å which is readily interpretable in terms of the molecular structure. Map interpretation was undertaken using the computer graphics O program, leading to over 90% of the protein being traced. Refinement of the model has been undertaken using a combination of RESTRAIN and XPLOR. Positional parameters have been refined for all atoms, but only side and main chain group isotropic thermal parameters for the individual residues, due to the resolution limit of 3.0 Å. The final model has some 1017 residues and a final R factor of 22.0% (start R = 34.6 %) and Free R = 28.6 %, (start Free R = 34.1%), for all reflections in the 12-3.0 Å range.