Crystal structure of ginkbilobin‐2 with homology to the extracellular domain of plant cysteine‐rich receptor‐like kinases

Fungi are an extremely diverse group of organisms with about 250,000 species and are found in all ecosystems.1 They are also proficient in colonization and infection of plants, and often cause harm to the host organisms. Most plants have evolved a variety of potent defense mechanisms against these pathogens, such as hypersensitive responses, reinforcement of cell walls, and synthesis of phytoalexins and antifungal proteins.2 To date, hundreds of antifungal proteins have been discovered in a wide variety of plants and are classified into the following groups: cyclophilins, defensins, pathogenesis-related proteins, ribosome-inactivating proteins, and so on.1 Recently, we discovered a novel antifungal protein, ginkbilobin-2 (Gnk2), in the endosperm of Ginkgo seeds.3 Gnk2 consists of 108 amino acids as a mature protein and inhibits the growth of phytopathogenic fungi such as Fusarium oxysporum.3 This antifungal protein shows no sequence similarity to other antifungal proteins.3 On the other hand, Gnk2 has considerable homology (∼85%) to embryo-abundant proteins (EAP) from the gymnosperms Picea abies and P. glauca, which suggests that Gnk2-like proteins are widely conserved in the seeds of gymnosperms. Plant EAP are expressed in the late stage of seed maturation and are involved in protection against environmental stresses such as drought.4 However, there are no reports suggesting that these proteins help defend against fungal pathogens. The sequence of Gnk2 is also 28–31% identical to the extracellular domain of cysteine-rich receptor-like kinases (CRK) from the angiosperm Arabidopsis. CRK has been classified as a member of the plant receptor-like kinases subfamily. There are more than 40 members of CRK in Arabidopsis and they contain 1–4 copies of domain 26 of unknown function (DUF26) with a C-X8-C-X2-C motif in their extracellular regions.5, 6 The CRK members are induced by pathogen infection and treatment with reactive oxygen species or salicylic acid7, 8 and are involved in the hypersensitive reaction, which is a typical system of programmed cell death.9 The C-X8-C-X2-C motif is completely conserved in the sequences of both Gnk2 and gymnosperms EAP. In addition, there are at least 60 genes in Arabidopsis encoding the cysteine-rich secreted proteins (CRSP) with the C-X8-C-X2-C motif. Therefore, the proteins with this motif are regarded as one of the largest protein superfamilies,5 although the role of the conserved motif remains unclear. Here, we report the crystal structure of Gnk2 with DUF26 and provide the first evidence of a structural role of the C-X8-C-X2-C motif. Purification, crystallization, and data collection of the native Gnk2 and the selenomethionine derivative of the recombinant Gnk2 (SeMet-rGnk2) crystals were performed as previously reported.10 The positions of four selenium atoms were detected in the asymmetric unit using the program SHELXD,11 and the phasing was performed by the MAD method using the program SHELXE.12 The initial model was automatically built using the program RESOLVE,13 and then refinement and water-addition procedures were manually performed with multiple cycles of the programs REFMAC514 and XtalView.15 Finally, the twin refinement of the model was conducted using the program PHENIX.16 Sequence and structure similarity searches were conducted using the BLAST services at the NCBI and DALI server (http://www.embl-ebi.ac.uk/dali/), respectively. Amino acid sequences were aligned and illustrated using the programs CLUSTAL W17 and ESPript,18 respectively. Computer graphic representations were prepared using the software GRASP19 and PyMOL (http://pymol.source forge.net). X-ray diffraction data were collected from the native Gnk2 crystal at 2.38 Å and from the SeMet-rGnk2 crystal at 2.79 Å resolution. These crystals were revealed to belong to the primitive cubic space group P213 as a merohedral twin with a twin fraction of 0.45. The crystal structure of Gnk2 was determined at 2.38 Å resolution by the MAD method using the twinned data and the final model was refined with the twinning operator of “l, −k, h.” Each Gnk2 crystal contained four molecules in an asymmetric unit, and the solvent content of the native crystal was 76.6%. The final model contained four sets of Ala1-Phe108 and 482 water molecules with a Rwork of 20.7% and Rfree of 23.2%. PROCHECK analysis20 revealed that 86.7% of all the residues were in the most favored region and the remaining residues were in the allowed region of the Ramachandran plot.21 The data collection and refinement statistics are summarized in Table I. The atomic coordinates and structure factors of Gnk2 have been deposited in the Protein Data Bank (PDB) under accession code 3A2E. The tertiary structure of Gnk2 was composed of two α-helices and a five-stranded β-sheet, which formed a compact single-domain architecture with an α+β-fold [Fig. 1(A)]. In the β-sheet, the order of the strands from one edge to the other was 1-4-5-3-2, and all the strands were in antiparallel orientations. The C-terminal half of Gnk2 (residues 56–108) corresponds to DUF26, which was composed of three β-strands (S3, S4, and S5) and the second α-helix (H2) [Fig. 1(A)]. DUF26 is generally referred to as a domain unit in CRK (see Fig. 2)22; however, Gnk2 revealed a compact single-domain architecture, and DUF26 forms the core region surrounded by the N-terminal half [Fig. 1(A)]. Here, we propose to define the Gnk2-homologous sequence as a new domain unit in CRSP or the extracellular regions of CRK. Overall Gnk2 structure with three disulfide bridges. (A) Ribbon diagram of the Gnk2 structure with two α-helices (magenta and red) and a five-stranded antiparallel β-sheet (blue and cyan). The DUF26 region on the Gnk2 molcule corresponds to the C-terminal side starting at the broken line, which is colored blue, gray, and red. (B) All the disulfide bridges between the H2 helix and the β-sheet. The disulfide bridges are shown as stick models with the yellow stick indicating a covalent bond between two sulfurs. The pink region of the H2 helix shows a kink near the disulfide bridge of Cys10-Cys85. (C) Electrostatic surface potential map of Gnk2. The charged areas are colored from blue (+10 kT/e) to red (−10 kT/e), and arginine residues are labeled with their residue numbers. Structure-based sequence alignment of Gnk2 and the Gnk2-homologous regions of the proteins with the C-X8-C-X2-C motif. Strictly conserved residues are shaded in red, and well-conserved ones are indicated with red letters. The secondary structural elements of Gnk2 are depicted at the top of the alignment. The signal-peptide sequence of Gnk2 is indicated with blue letters. The green-boxed sequences correspond to the motif, and the DUF26 sequences are underlined in orange. The following sequences are used for comparison: EAP_Pa, EAP from Picea abies; EAP_Pg, EAP from P. glauca; CRK_At, CRK from Arabidopsis thaliana; CRSP_At, CRSP from A. thaliana. Cysteine residues formed three intramolecular disulfide bridges: Cys10-Cys85, Cys62-Cys71, and Cys74-Cys99. These bridges were connected by three β-strands (S1, S3, and S5) to the H2 helix, and Cys10-Cys85 caused a kink at the position of Ile82 within the H2 structure [Fig. 1(B)]. The H2 helix also seems to be packed against the β-sheet using the three disulfide bridges. For the first time, this work demonstrates the pattern of three disulfide bridges in the C-X8-C-X2-C motif conserved in CRSP and the extracellular regions of CRK [Figs. 1(B) and 2]. In a previous article, the reaction solution of Gnk2 was reported to show gelation by performic-acid oxidation.3 This indicates that the disulfide bridges between the H2 helix and β-sheet largely contribute to the structural stability of Gnk2. Presumably, the absence of the disulfide bridges could result in the conformational change of the H2 helix, leading to the oligomerization and insolubility of the oxidized molecule. A structural-homology search by the DALI server showed that there are 22 proteins homologous to Gnk2, with Z-values of 2.0–3.0, where the Z-value represents the degree of structural similarity.23 However, the 22proteins were not related to any antifungal protein; for instance, the protein with the highest Z-value of 3.0 was a transcriptional coactivator for steroid receptors (r.m.s.d., 4.3 Å; number of Cα atoms, 68; sequence identity, 9%; PDB code, 1OJ5). This indicates that the Gnk2 possessed a novel fold and an apparently distinctive structure when compared with other antifungal proteins. With regard to antifungal proteins, both defensin and Gnk2 are cysteine-rich proteins with an α+β-fold and three or four disulfide bridges.24 The positively charged surface of defensin contributes to the interaction with negatively charged phospholipids.25 The first α-helix (H1) of Gnk2 included four arginine residues, Arg23, Arg26, Arg33, and Arg47. These guanidinium groups had no ionic interactions or formed salt bridges with any other group and formed a positively charged surface around the H1 helix [Fig. 1(C)]. We speculate that the antifungal activity of Gnk2 may require an association between its positively charged surface and the negatively charged phospholipids and/or phosphomannan on the fungal-cell surface. We would like to thank the scientists and staff at the Photon Factory. The synchrotron-radiation experiments were conducted at AR-NW12 and BL-17A at the Photon Factory, Tsukuba, Japan (Proposal Nos. 2006S2-006 and 2007G-163). We also thank the National BioResource Project (NIG, Japan) for the gift of E. coli (JM109 pDsbABCD1).