Phospholipids are major components of all living cells. They are synthesised and broken down by a diverse family of enzymes known as phospholipases. These are subdivided into classes A-D depending on where they attack the phospholipid molecule. Many of the products are used as signalling molecules, which enable cells to react to internal and external stimuli. As part of a long term structural study of phospholipases the structure of the 56 Kda Phospholipase D from Streptomyces sp. strain PMF (PLDPMF) has been solved at the ESRF [1]. Diffraction data to 1.4 Å resolution for the enzyme and several potential heavy atom derivatives were collected on BM1 (the Swiss Norwegian Beamline). None of the derivatives proved to be suitable to solve the structure but the presence of a single highly occupied tungstate ion was detected in one of them. This enabled a 4-wavelength MAD experiment on BM14 which rapidly lead to solution of the structure at 1.9 Å including an almost complete amino acid sequence. After solvent flattening, histogram matching and phase extension, the structure was refined to convergence using SHELXL with the 1.4 Å BM1 data [2].

A ribbon representation of PLDPMF is shown in Figure 18. The protein backbone goes from N to C in this figure and is coloured through the rainbow so that one can see that it is folded into two topologically similar regions, so-called "domains", which are related by a two-fold rotation axis. The active site of the enzyme, occupied by a phosphate group in the figure, lies on this axis with a histidine, an asparagine and a lysine residue from each domain. The sequence pattern for these residues in the two domains is, in both cases HXKX4DX9N.

This is the first structure of a phospholipase D. However it provides structural data for an extensive family of proteins with quite diverse biochemical activity. The "PLD-family" was identified by Ponting et al. [3] by searching protein sequence data bases for characteristic strings of amino acids. One of these strings was the HXKX4DX9N pattern which we have now found to occur twice in the active site of PLDPMF. The PLD family also includes a 16 KDa endonuclease, nuc, from Salmonella typhimurium. Although this enzyme is much smaller than PLDPMF and only contains a single copy of this sequence string, a recent crystal structure [4] has shown that nuc actually forms a dimer with an active site very similar to that in PLD.

As a consequence of the way the structure was determined it was found that one of the amino acids in the active site lies in two different orientations depending on whether enzyme activity is blocked by e.g. phosphate or tungstate or is available for catalysis. This lead to a series of experiments on ID14-4, where crystals were soaked with a soluble phospholipid for different times before freezing and data collection. This has given a complete and detailed picture of the sequence of events, which occur when PLDPMF breaks down its substrate into a so-called phosphatidic acid and choline [5]. In one of these experiments the chemical reaction was trapped in a so-called transition state. The experiments also caused some surprise since it appears that the enzyme is also able to hydrolyse its own phosphatidic acid product further, but that in doing so probably deactivates itself.

References
[1] I. Leiros, E. Hough, P. D'Arrigo, G. Carrea, G. Pedrocchi-Fantoni, F. Secundo, S. Servi, Acta Cryst. D, 56, 466-468 (2000).
[2] I. Leiros, F. Secundo, C. Zambonelli, S. Servi, E. Hough, Structure with Folding & Design, 8, 655-667 (2000).
[3] C.P. Ponting, I.D. Kerr, Protein Sci, 5, 914-922 (1996).
[4] J.A. Stuckey, J.E. Dixon, Nature Struct. Biol., 6, 278-284 (1999).
[5] I. Leiros, S. McSweeney, E. Hough, To be published.

Authors
E. Hough (a), S. McSweeney (b), I. Leiros (a).

(a) University of Tromso (Norway)
(b) EMBL ­ Grenoble Outstation (France)