Correspondence e-mail:

MolProbity structure validation will diagnose most local errors in macromolecular crystal structures and help to guide their correction.


Introduction
Lipid-transfer proteins (LTPs) are ubiquitous lipid-binding proteins from plants. Two multigenic families have been identified, LTP1 and LTP2. Both LTP1 and LTP2 display low or no amino-acid sequence identity except for a common motif of eight cysteines involved in four disulfide bridges (Fig. 1). This motif is now considered to be the LTP signature and reflects a common helical fold stabilized by four helices. LTPs have a large central hydrophobic tunnel or cavity that allows the binding of different types of lipids and hydrophobic molecules. In Arabidopsis thaliana, more than 20 genes encoding LTPs have been highlighted. They are expressed or induced during a wide range of biotic or abiotic stress. These proteins have been associated with such different physiological events that their precise functions still remain to be discovered (Douliez et al., 2000;Marion et al., 2004). However, most of the data point to a role of these proteins in the defences of plants against their microbial predators. In particular, it has been shown that LTP1s and oomycetous elicitins compete for a plasma membrane receptor involved in the hypersensitive response (HR) and systemic acquired resistance (SAR) in Nicotiana tabacum (Buhot et al., 2001;Blein et al., 2002). This is strengthened by the observation that a mutant of A. thaliana defective in a gene encoding a putative LTP2, i.e. DIR1, was unable to develop SAR (Maldonado et al., 2002). Therefore, all these studies suggest that in planta LTPs are probably involved in the signalling of plant defence mechanisms. DIR1 (defective in induced resistance 1) is not the signalling molecule stricto sensu, but could be involved in the transport of lipids that are the effective SAR signals (Maldonado et al., 2002). Therefore, to understand the relationship between the biological function of DIR1 and its lipid-binding properties, we have crystallized a protein-lipid complex of DIR1.

Preparation and purification
The mature DIR1 protein (gene accession No. At5g48485) from A. thaliana was cloned into pPICZ vector (Invitrogen) and expressed in the methylotrophic yeast Pichia pastoris. The yeast culture supernatant was mostly constituted of the native DIR1 # 2006 International Union of Crystallography All rights reserved Figure 1 LTP2 alignments. The conserved cysteine residues are boxed. The numbering is given with respect to DIR1.
protein. It was first centrifuged and then purified using the following three-step method.
Firstly, the Pichia supernatant was concentrated and desalted using adsorption chromatography (XAD7 resin, Amberlite). Secondly, the protein was eluted using an acetonitrile gradient (50% acetonitrile, 0.1% TFA). Thirdly, the resulting material was loaded onto a High Prep SP-XL column (Amersham) pre-equilibrated with 50 mM sodium acetate pH 3.5 (elution took place using a 1 M NaCl gradient at pH 3.5). The native protein was finally purified using reverse-phase chromatography. The homogeneity and the purity of the protein were checked by mass spectrometry.

Crystallization
The protein was first incubated with monostearoyl phospatidylcholine as a putative substrate in a ratio of 1:1.5 for 1 d at 277 K. Preliminary screenings were performed using a Cybi-Disk robot system with Hampton Research Crystal Screen I and II kits. 96-Condition plates were employed with a 100 ml reservoir and 0.5 ml drops composed of 50% reservoir solution and 50% protein-complex solution (20 mg ml À1 ). The best hits leading to crystals always contained zinc salts and PEG MME at acidic pH. Optimization of these conditions took place in standard Linbro plates. The final optimized conditions for routine preparation were 0.08 M acetate buffer pH 6.0 containing 0.02 M ZnSO 4 and 20-25% PEG 600 monomethyl ether as the reservoir solution and a hanging drop made up of 1 ml of protein at 20 mg ml À1 in 0.02 M ZnSO 4 pH 6.0 mixed with 1 ml reservoir solution. Crystals grow to their final size (Fig. 2) in about a month.

Data collection
Data collection took place at ESRF, Grenoble on beamline ID-29 using an ADSC CCD detector. Crystals were frozen in liquid nitrogen. No cryoprotectant was used owing to the high polyethylene glycol content. Three different strategies were applied. An initial highly redundant data collection (400 rotation) was recorded at 1.9 Å in order to exploit the anomalous signal of sulfur, which was present at a high level (eight cysteines forming four disulfide bridges plus one methionine in 77 residues). A second round of three data collections took place using the same crystal at the Zn K edge (for MAD techniques) and a final data collection was performed at the highest possible resolution with a fresh crystal at a shorter wavelength. The data collections and their diffraction statistics are reported in Table 1. In all cases, the diffraction extends to a limit of 1.6-1.9 Å ; however, this was associated with a high mosaicity.
All data were processed using MOSFLM, merged with SCALA and finally put on an absolute scale with TRUNCATE, all of which are available from the CCP4 program suite (Collaborative Computational Project, Number 4, 1994). The unit-cell parameters (a = 28.23, b = 48.15, c = 55.01 Å ) and volume (V = 74 786 Å 3 ) correspond to a Matthews coefficient of 2.32 Å 3 Da À1 , which accounts for one molecule in the asymmetric unit and a solvent content of 47%.
Preliminary attempts to solve the structure by molecular replacement using the wheat (PDB code 1tuk; Hoh et al., 2005), rice (PDB code 1l6h; Samuel et al., 2002)

Figure 2
Crystals of DIR1. Maximum size is 0.2 mm. crystal structure of DIR1 and are therefore attempting to solve the structure using the MAD data at the Zn K edge.
We thank the staff of the ID-19 beamline, ESRF for beam time access.