View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by Elsevier - Publisher Connector Structure, Vol. 11, 703–713, June, 2003, 2003 Elsevier Science Ltd. All rights reserved. DOI 10.1016/S0969-2126(03)00094-7 The Crystal Structure of Choline Kinase Reveals a Eukaryotic Protein Kinase Fold Daniel Peisach, Patricia Gee, Claudia Kent, berg and Kornberg, 1953). The enzyme from several and Zhaohui Xu* mammalian sources (Ishidate et al., 1984; Porter and Department of Biological Chemistry Kent, 1990; Uchida and Yamashita, 1990) as well as University of Michigan Medical School Saccharomyces cerevisiae (Kim et al., 1998) has been 1301 East Catherine Road subsequently purified to homogeneity. Most choline ki- Ann Arbor, Michigan 48109 nases utilize both choline and ethanolamine as sub- strates to accept phosphotransfer from ATP, although Km values for choline are lower than those for ethanol- Summary amine. Enzymes that are specific for ethanolamine have also been reported (Lykidis et al., 2001; Pavlidis et al., Choline kinase catalyzes the ATP-dependent phos- 1994; Uchida, 1997). Genes and cDNAs for bona fide phorylation of choline, the first committed step in the choline/ethanolamine kinases have been isolated from CDP-choline pathway for the biosynthesis of phospha- mammals (Aoyama et al., 1998, 2000; Uchida and Ya- tidylcholine. The 2.0 A˚ crystal structure of a choline mashita, 1992), plants (Al-Malki et al., 2000; Monks et kinase from C. elegans (CKA-2) reveals that the en- al., 1996), and yeast (Hosaka et al., 1989), and recent zyme is a homodimeric protein with each monomer genome sequencing efforts have identified many more organized into a two-domain fold. The structure is putative choline and/or ethanolamine kinases. Align- remarkably similar to those of protein kinases and ment of the protein sequences of choline kinase from aminoglycoside phosphotransferases, despite no sig- several species reveals several highly conserved re- nificant similarity in amino acid sequence. Compari- gions, including two clusters of conserved residues (Fig- sons to the structures of other kinases suggest that ure 1). The first, also known as the Brenner’s motif, has ATP binds to CKA-2 in a pocket formed by highly con- been identified in many enzymes catalyzing phospho- served and catalytically important residues. In addi- transfer reactions, including protein kinases (Aoyama et tion, a choline binding site is proposed to be near the al., 2000; Brenner, 1987). The second conserved cluster ATP binding pocket and formed by several structurally appears to be specific for members of the choline/etha- flexible loops. nolamine kinase family (the choline kinase motif), raising the possibility that it might be important for binding choline or ethanolamine. Introduction Multiple choline kinase isozymes are found in many organisms, as indicated by both biochemical (Ishidate Phosphatidylcholine is a major structural component of et al., 1982, 1985; Porter and Kent, 1990) and genetic eukaryotic cellular membranes and serum lipoproteins evidence (Aoyama et al., 2000; Monks et al., 1996; and serves as a precursor for the production of lipid Uchida, 1994). It is possible that induction of expression second messengers (Exton, 2000). The predominant of different isozymes is a mechanism for regulating cho- pathway for the biosynthesis of phosphatidylcholine in line kinase activity, but roles for the various isoforms animals and plants is the three-step CDP-choline path- have not been well defined. Analysis of the C. elegans way, whereby choline is first converted into phospho- genome has identified seven choline kinase-like genes, choline, which then reacts with CTP to form CDP-cho- which fall into three groups based on similarity in se- line. Finally, the phosphocholine moiety is transferred to quence to mammalian choline kinases; group A is most diacylglycerol to produce phosphatidylcholine. Choline similar while groups B and C are less so (Gee and Kent, kinase catalyzes the first step of the pathway, the ATP- 2003). Four of the seven genes (two from the A group dependent phosphorylation of choline, and has been and two from the B group) have recently been cloned reported to regulate the CDP-choline pathway (Kent, and expressed in insect cell lines. All of them have cho- 1990). In mammals, phosphocholine has been reported line kinase activity; those that most closely resemble to be mitogenic (Cuadrado et al., 1993), and inhibitors the mammalian choline kinases are the most active. An of choline kinase are being investigated as therapeutic example of this is isoform A-2 (CKA-2). It exists as a agents in cancer research (Hernandez-Alcoceba et al., dimer composed of two identical subunits of 429 resi- 1997). In addition to its importance in the biosynthesis dues each. Biochemical studies have shown that this of phosphatidylcholine, choline kinase functions in the enzyme has Km values of 1.6 mM and 2.4 mM for choline biosynthesis of cell wall-associated phosphocholine in Ϫ1 and ATP, respectively, and a kcat of 74 s . The enzyme bacteria (Bean and Tomasz, 1977; Poxton and Leak, requires magnesium for activity, has a very alkaline pH 1977; Tuomanen, 1999). For example, the necessity of optimum, and is much more active with choline as a phosphocholine to S. pneumoniae cell wall structure substrate than with ethanolamine. We have begun to (Tuomanen, 1999) renders its choline kinase as a possi- use this enzyme to study the functions of the various ble antibacterial target. choline kinase-like genes in C. elegans, as well as to A partially purified preparation of choline kinase from provide a model for structure-function studies of the brewer’s yeast was first studied fifty years ago (Witten- broader choline kinase family. In the present study, we determined the three-dimen- *Correspondence: [email protected] sional structure of CKA-2 in order to understand the Structure 704 Figure 1. Sequence Alignment of Choline/Ethanolamine Kinases The sequences of C. elegans choline kinase CKA-2, human choline/ethanolamine kinase (CK/EK), mouse choline/ethanolamine kinase, human ethanolamine kinase, Drosophila ethanolamine kinase, and S. cerevisiae choline kinase were aligned with the program ClustalW (Higgins et al., 1994). Invariant residues are darkly shaded and outlined. Similar, but not identical, residues are lightly shaded and outlined. Secondary structure elements are indicated above the sequence: ␣ helices, rectangles; ␤ strands, arrows; other elements, solid lines; structurally unobserved residues at the N terminus and midsequence, dashed lines. These elements are colored on the basis of their locations in the sequence (see text for detail). For purpose of clarity, sequences in non-CKA-2 proteins where no homology is identified with CKA-2 have been replaced by bracketed numbers indicating the number of residues omitted from the display. Three highly conserved residues from the proposed ATP binding site are indicated with red letters. molecular mechanisms of its catalysis and substrate strate binding sites. It is noteworthy that CKA-2 has selectivity for choline over ethanolamine and to estab- remarkable structural similarity to the eukaryotic protein lish a foundation for structure-function relationships. We kinase family and the bacterial aminoglycoside phos- report here the general features of this choline kinase photransferase family, despite limited sequence similar- and detailed structural information on its active and sub- ity. Our comparison suggests that choline kinase could The Crystal Structure of Choline Kinase 705 use a similar mechanism to those of other phosphotrans- and bAP in this region is striking. For example, despite ferases in catalysis. This work represents the first struc- the low sequence identity between CKA-2 and PKA, the tural analysis of a choline/ethanolamine kinase. rms deviation for the structurally conserved main chain atoms in the N-terminal domain is only 1.7 A˚ . The N-terminal domain is linked to the C-terminal do- Results and Discussion main through a short strand (strand 6) bordered by two 90Њ turns (Figures 1 and 2A, yellow). The C-terminal do- Structure Determination main can be subdivided into three regions: a central The CKA-2 protein used in the structure determination core subdomain (Figures 1 and 2A, green), an insertion was overexpressed and purified from E. coli. The bacte- subdomain (Figures 1 and 2A, blue), and a C-terminal rially expressed protein had a nearly identical catalytic subdomain (Figures 1 and 2A, magenta). The central profile to the one expressed in eukaryotic (insect) cells core consists of three helices (C, D, and G) and a long and was crystallized in two different, yet related, crystal hairpin-shaped loop that incorporates two short stretches forms. Crystals grown from the selenomethionine-sub- of antiparallel ␤ strands (9, 10, 11 and 12). Strands 10 stituted protein belong to the orthorhombic space group and 11, together with strand 6, form a small antiparallel P2 2 2, and those from the native protein belong to the 1 1 ␤ sheet on the right side of the molecule (Figure 2A). monoclinic space group P2 . The two have nearly identi- 1 This long hairpin loop contains many of the conserved, cal crystal packing, with the crystallographic two-fold c functionally important residues. For example, strands axis in the P2 2 2 crystals becoming a noncrystallo- 1 1 9–10 house the Brenner’s motif, and strands 11–12 graphic two-fold axis in the P2 crystals. The structure of 1 house the choline kinase motif (Figure 2B). As will be CKA-2 was first determined by the multiple-wavelength discussed in following sections, these sequences are anomalous diffraction (MAD) method with selenium most likely involved in binding substrates and per- atoms in selenomethionine as the anomalous scattering forming catalysis. In addition, strands 10 and 11 are atoms and later refined against a 2.0 A˚ resolution native connected by a long stretch of sequence (residues 265– data set.