Prokaryotic Type II and Type III Pantothenate Kinases: the Same Monomer Fold Creates Dimers with Distinct Catalytic Properties

Home , PANK2 (gene), Pantothenate kinase

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Structure 14, 1251–1261, August 2006 ª2006 Elsevier Ltd All rights reserved DOI 10.1016/j.str.2006.06.008 Prokaryotic Type II and Type III Pantothenate Kinases: The Same Monomer Fold Creates Dimers with Distinct Catalytic Properties

Bum Soo Hong,1,6 Mi Kyung Yun,3,6 Yong-Mei Zhang,4 from their bacterial counterparts (Calder et al., 1999), Shigeru Chohnan,4,7 Charles O. Rock,4,5 and they are feedback inhibited by CoA or its thioesters Stephen W. White,3,5 Suzanne Jackowski,4,5 to achieve a tight and tissue-specific regulation of the Hee-Won Park,1,2 and Roberta Leonardi4,* cofactor level (Rock et al., 2000, 2002; Zhang et al., 1 Structural Genomics Consortium 2005). 2 Department of Pharmacology Bacteria possess three types of pantothenate kinases University of Toronto (CoaAs) that differ in sequence, regulation, substrate af- Toronto, Ontario M5G 1L5 finity, and specificity (Brand and Strauss, 2005; Leonardi Canada et al., 2005a; Vallari et al., 1987). The type I CoaA is 3 Department of Structural Biology widely distributed, and the prototypical member is the 4 Department of Infectious Diseases Escherichia coli CoaA (EcCoaA), which is encoded by St. Jude Children’s Research Hospital the coaA gene (Song and Jackowski, 1992, 1994) and Memphis, Tennessee 38105 is regulated through feedback inhibition by CoA and 5 Department of Molecular Sciences CoA thioesters (Vallari et al., 1987). EcCoaA is a dimer University of Tennessee Health Science Center of identical subunits and is structurally classified as Memphis, Tennessee 38163 a P loop kinase containing a Walker A motif (Ivey et al., 2004; Yun et al., 2000). The pantothenate kinase from Staphylococcus aureus (SaCoaA) is the prototypical Summary type II enzyme, and it is refractory to feedback inhibition by CoA or its thioesters (Leonardi et al., 2005a). This lack Three distinct isoforms of pantothenate kinase (CoaA) of regulation allows the accumulation of millimolar con- in bacteria catalyze the first step in coenzyme A bio- centrations of CoA, which is the major intracellular thiol synthesis. The structures of the type II (Staphylococ- in this organism (delCardayre et al., 1998). There is par- cus aureus, SaCoaA) and type III (Pseudomonas aeru- ticular interest in the type II SaCoaA protein because it is ginosa, PaCoaA) enzymes reveal that they assemble related to the four eukaryotic isoforms of pantothenate nearly identical subunits with actin-like folds into di- kinase, PanK1a, PanK1b, PanK2, and PanK3 (Cheek mers that exhibit distinct biochemical properties. et al., 2005). Alignment of the human PanK2 and SaCoaA PaCoaA has a fully enclosed pantothenate binding sequences (Figure 1A) reveals that the region of greatest pocket and requires a monovalent cation to weakly similarity corresponds to the conserved phosphate bind ATP in an open cavity that does not interact binding P loop DxGG(T,S)T(S,G)xxK(R,C) motif charac- with the adenine nucleotide. Pantothenate binds to teristic of members of the acetate and sugar kinases/ an open pocket in SaCoaA that strongly binds ATP Hsc70/actin (ASKHA) family (Bork et al., 1992; Kabsch by using a classical P loop architecture coupled with and Holmes, 1995). Finally, a number of bacteria, includ- specific interactions with the adenine moiety. The Pa- ing pathogens such as Helicobacter pylori and Pseudo- CoaAPan binary complex explains the resistance of monas aeruginosa, possess a type III enzyme. The type bacteria possessing this isoform to the pantothena- III CoaA from H. pylori, HpCoaA, is not regulated by mide antibiotics, and the similarity between SaCoaA CoA or its thioesters, and, unlike the type I and II CoaAs and human pantothenate kinase 2 explains the molec- that have similar Kms for the substrates in the micro- ular basis for the development of the neurodegenera- molar range (Leonardi et al., 2005a; Song and Jacko- tive phenotype in three mutations in the human wski, 1994), it exhibits an unusually high Km for ATP protein. (w10 mM) (Brand and Strauss, 2005). The absence of any obvious sequence similarity between the three types of bacterial CoaAs, and their strik- Introduction ingly different biochemical properties, prompted us to investigate the structural basis of this diversity. Our Coenzyme A (CoA) is an indispensable cofactor in all bi- study reports the crystal structures of the type II ological systems, where it functions as the major acyl SaCoaAAMP-PNP complex, the type III PaCoaA apo- group carrier (Leonardi et al., 2005b). The universal enzyme, and the PaCoaAPan complex. Both enzymes CoA biosynthetic pathway consists of five enzymatic share the same actin-like monomer fold, but they asso- steps, the first of which is the phosphorylation of panto- ciate to form two very different dimeric architectures. thenate (Pan), or vitamin B5, catalyzed by pantothenate Furthermore, by combining the structural information kinase. Pantothenate kinase is the key regulatory step in on how SaCoaA and PaCoaA bind the phosphoryl donor CoA biosynthesis (Leonardi et al., 2005b). The eukary- and Pan, respectively, we can assemble an accurate otic pantothenate kinase isoforms (PanKs) are distinct picture of the catalytic ternary complex for each enzyme. These models provide a structural framework for understanding the substrate specificities of each en- *Correspondence: [email protected] zyme and their distinct catalytic mechanisms. Both the 6 These authors contributed equally to this work. 7 Present address: Department of Bioresource Science, College of type I and type II CoaAs recognize the pantothenamide Agriculture, Ibaraki University, 3-21-1 Chu-ou, Ami, Ibaraki class of antimetabolites (Leonardi et al., 2005a; Strauss 300-0393, Japan. and Begley, 2002), whereas the type III CoaAs do not Structure 1252

see the Supplemental Data available with this article on- line). The monomer structures are very similar and com- prise two a/b domains, and each domain contains a mixed five-stranded b sheet surrounded by a helices (Figures 2A and 2C). Domain I consists of strands b1– b5 in the order 3-2-1-4-5 with strand b2 antiparallel to the others, and helices a1, a2, a3, and a8. Domain II contains strands b6–b10 in the order 8-7-6-9-10 with strand b7 antiparallel to the others, and helices a3, a4, a5, a6, a7, and a8. The folding topologies of each domain are identical and place them in the actin superfamily. They are tightly associated such that helices a3 and a8 are common to each domain at the domain interface, with the C-terminal a8 completing the domain I substructure. A structural similarity search of the protein database with the VAST program revealed that the fold of SaCoaA and PaCoaA places them in the ASKHA superfamily (Bork et al., 1992; Cheek et al., 2002; Kabsch and Holmes, 1995). Members of this family have the same relative orientations of domains I and II. The classic P loops in the kinase fold are between the first and second b strands of each domain (strands b1 and b2 of domain I and strands b6 and b7 of domain II), and these meet at the center of the domain interface. The domain I P loop has the classical signature sequence, whereas domain II can be considered a ‘‘pseudo kinase’’ domain. Domain II is larger than domain I by virtue of two additional helices, a4 and a5, and an extended a6 compared to its counterpart, a1(Figures 2A and 2C). Both proteins crystallized as a dimer in the asymmet- Figure 1. Alignment of Type II and Type III CoaAs ric unit, and the monomers of this dimer are related by (A) Alignment of the primary sequences of SaCoaA and human a noncrystallographic 2-fold rotation axis. The dimen- PanK2. Residues highlighted in red were investigated by site-di- sions of the dimers are consistent with the hydrody- rected mutagenesis, and those indicated with an asterisk (*) are con- namic data on the two soluble proteins (Figure S1)(Leo- served between SaCoaA and human PanK2 and are mutated in nardi et al., 2005a). Although the SaCoaA and PaCoaA PKAN patients. The residues in the dimer interface conserved between the two proteins are indicated with a ‘‘+.’’ monomers have very similar structures, they each con- (B) Alignment of the primary sequences of the type III CoaAs from tain a loop region that is absent from the other enzyme, P. aeruginosa (NCBI gij15599475), B. subtilis (NCBI gij16077138, which results in their assembly into dimers with dis- corrected in GenBank acc. #AY912104), and H. pylori (NCBI tinctly different architectures (Figures 2B and 2D). Sa- gij15645481). Sequence identities are: PaCoaA/BsCoaA = 19%, CoaA contains a loop between helices a4 and a5 that PaCoaA/HpCoaA = 21%, HpCoaA/BsCoaA = 14%, and PaCoaA/ packs against a6(Figure 2B). Conversely, PaCoaA has SaCoaA = 12%. The residues in the dimer interface conserved in b a at least two of the three sequences are indicated with a ‘‘+.’’ a loop inserted between 5 and 3 that contains a short b ribbon that associates with the outside b8 strand of the domain II b sheet (Figure 2D). Although both loops con- accept these compounds as substrates (Brand and tribute to the monomer-monomer interface, they do so Strauss, 2005). The different dimeric structures of in different ways. The SaCoaA loop mainly interacts SaCoaA and PaCoaA explain the resistance of P. aerugi- with a40, whereas the PaCoaA loop interacts with the nosa to the pantothenamide class of drugs. Finally, the a50–a60 loop (the prime sign indicates an element in the structure of the SaCoaAAMP-PNP binary complex pro- other monomer). As a result, the long a6 helices within vides insights into the structure and catalytic mecha- the SaCoaA dimer interact extensively via a parallel nism of the mammalian pantothenate kinases. This is coiled coil (Figure 2B), whereas the same helices in the important because the human PanK2 isozyme is associ- PaCoaA dimer interact at a 70 angle and contribute ated with mutations that result in the progressive neuro- less to the dimer interface (Figure 2D). degenerative disorder known as PKAN (pantothenate ki- In both the SaCoaA and PaCoaA crystal structures, nase-associated neurodegeneration) (Zhou et al., 2001), the dimer is the asymmetric unit, and their different ar- and we can now provide insights into the structural chitectures are unlikely to be the result of crystallization basis for the enzyme inactivation by these mutations. artifacts. This is supported by an analysis of the conser- vation of the residues that mediate the different dimer in- Results and Discussion terfaces. The PaCoaA interface buries 4243 A˚ 2 of surface area and involves 53 residues that are highly conserved Overall Structures of SaCoaA and PaCoaA (Figure 1B). A ‘‘fingerprint’’ of the interface is provided The crystal structures of SaCoaA and PaCoaA were both by four conserved small residues that occupy tight determined by MAD phasing and were subsequently re- spaces between the monomers, glycines 139, 187, and fined to 2.05 and 2.2 A˚ , respectively (Tables S1 and S2; 194, and Ala183. The latter three are within helix a6, Type II and Type III Pantothenate Kinases 1253

Figure 2. The Overall Structures of SaCoaA and PaCoaA (A–D) The (A) SaCoaA and (C) PaCoaA monomers are shown in identical orientations to highlight their structural similarities and their different packing interactions with their (B and D) dimeric partners. The secondary structure elements are labeled in monomer A, and the N and C termini are indicated. Helix a6 is shown in red, and the loops that distinguish the two monomers are shown in blue. Pan molecules are modeled according to the PaCoaA structure and are shown with cyan carbons. The AMP-PNPMg2+ molecules modeled according to the closely related SaCoaA structure are shown with gray carbons. Dashed lines represent residues that are not visible in the electron density. (A) The monomer structure of SaCoaA. (B) In the SaCoaA dimer, helices a6 pack in a parallel coiled-coil arrangement. (C) The monomer structure of PaCoaA. (D) The PaCoaA dimer forms with the central helix a6 packing against its partner at an angle of 70. and Gly194 packs against its counterpart where a6 and (Kachmer and Boyer, 1953; Oria-Hernandez et al., a60 cross each other. In SaCoaA, the dimer interface bur- 2005; Suelter, 1974). The majority of the enzymes requir- ˚ 2 + + + ies 2962 A and involves 39 residues. These interface ing K are also activated by NH4 and Rb but are poorly residues are the most highly conserved in the human activated by cations with smaller radii, such as Na+ and PanK2 primary sequence outside of the active site Li+. PaCoaA showed a similar behavior with a strong + (Figure 1A), and this ﬁnding supports both the dimer preference for NH4 , which induced a 5-fold higher acti- structure and the homology between SaCoaA and vation than either K+ or Rb+ (Figure 3A). The type III en- PanK2. The unique loop regions in SaCoaA and PaCoaA zymes from B. subtilis (BsCoaA) and H. pylori (HpCoaA) + also support their different dimer architectures. Thus, were also strictly dependent on the presence of NH4 or the SaCoaA loop cannot be accommodated in the Pa- K+ for activity (Figure 3B). These biochemical properties CoaA dimer, and the PaCoaA loop contains a highly distinguish the type III pantothenate kinases from the conserved Tyr92 that interacts with Pan (see below) type I and II CoaAs. and occupies a pocket at the interface. The kinetic parameters of PaCoaA for ATP and pantothenate were determined in the presence and absence + PaCoaA Requires a Monovalent Cation for Activity of NH4 to investigate the activating effect of this mono- + Type I and type II CoaAs from E. coli and S. aureus, re- valent cation on the enzyme. Since the omission of NH4 spectively, catalyze the phosphorylation of pantothe- would have completely abolished the enzymatic activity, nate in the presence of ATP and MgCl2 (Ivey et al., preventing any comparison, KCl was included in both 2004; Leonardi et al., 2005a; Song and Jackowski, sets of conditions. The addition of NH4Cl had no effect 1994). However, PaCoaA was inactive under the same on the kinetic parameters for pantothenate (Fig- assay conditions, suggesting that an additional cofactor ure S2A). Conversely, the ammonium affected both the was required. The enzymatic activity of a type III CoaA Km and the kcat for ATP, with an overall 4- to 5-fold in- was previously measured by using a coupled assay crease in the speciﬁcity constant for this substrate, + that included phosphoenolpyruvate, NADH, and KCl which accounts for the stronger activation by NH4 + (Brand and Strauss, 2005). We therefore tested the ef- with respect to K (Figure 3). The Km for ATP remained + fect of these components on the PaCoaA activity and in the millimolar range even in the presence of NH4 , found that KCl activated the enzyme. The optimal KCl and this suggests that ATP may not be the preferred concentration was 60–120 mM (Figure 3A). The require- substrate. The activity of PaCoaA was therefore as- ment for a monovalent cation is common to several en- sayed in the presence of alternative nucleotides and zymes that catalyze phosphoryl transfer reactions deoxynucleotides. PaCoaA utilized a wide range of Structure 1254

2003). None of these substrates supported pantothenate kinase activity.

The ATP Binding Sites In the SaCoaAAMP-PNP complex, the ATP analog binds within the interdomain cleft of each SaCoaA monomer (Figure 2A), and the triphosphate moiety binds at the convergence of the P loop in domain I and the ‘‘pseudo P loop’’ in domain II. Interactions with the conserved P loop residues are typical for a kinase, and the key interactions are shown in Figure 4A. The main chain nitrogen of Gly9 is hydrogen bonded to the nonbridging oxygen of the g-phosphate, and it also interacts with the bridging oxygen of the b-phosphate. The NZ nitrogen of Lys13 forms hydrogen bonds with the oxygens of both the a- and b-phosphates of AMP-PNP. A Mg2+ ion is coordinated by the b- and g-phosphates of the AMP-PNP and several water molecules, and Asp6 interacts with the magnesium through one of these coordinating water molecules. Thr10 is adjacent to the b-phosphate, and its main chain nitrogen is hydrogen bonded to the nonbridging oxygen of the b-phosphate (data not shown). As regards interactions with the pseudo P loop, the OG1 oxygen of Thr99 is hydrogen bonded to the bridging nitrogen of the b- and g-phosphates, and its main chain nitrogen forms a hydrogen bond to the bridging oxygen of the a- and b-phosphates. The main chain nitrogen of Gly100 forms a water-mediated hydrogen bond with a nonbridging oxygen of the g-phosphate, and Leu11 and Ser225 provide two additional main chain nitrogens for interactions with the a- and b-phosphates. We were not successful in obtaining the structure of the PaCoaAnucleotide complex, perhaps due to the high Km of PaCoaA for ATP, but the crystal structure of the SaCoaAAMP-PNP complex was used to generate a model (Figure 4B). Asp6, Gly8, and Lys13 in the PaCoaA P loop are positioned to form interactions equivalent to those observed in SaCoaA, although Lys13 interacts with the carbonyl group of Pro234 and the side chain of Asp235 in the absence of nucleotide. Figure 3. Monovalent Cation Requirement for Type III CoaAs The classic P loop in kinases has a conserved (A) The activity of PaCoaA was assayed in reaction mixtures contain- DxGG(T,S)T(S,G)xxK(R,C) motif that is present but not ing 300 ng enzyme, 100 mM HEPES (pH 7.5), 45 mM D-[1-14C]panto- totally conserved in SaCoaA and PaCoaA (DAGGTLIK thenate, 250 mM ATP, 10 mM MgCl2, 0.1 mg/ml g-globulin, and in- and DCGNSLIK, respectively). We therefore performed creasing concentrations of LiCl (closed circle), NaCl (open circle), mutagenesis studies to verify the roles of these nucleo- KCl (closed square), RbCl (open square), and NH4Cl (closed trian- tide binding residues that were revealed in the SaCoaA- gle), as indicated. AMP-PNP crystal structure and predicted in the Pa- (B) Comparison between the activating effect of increasing concen- CoaAATP model. The SaCoaA[T10A] mutant was trations of KCl (closed square, open square) and NH4Cl (closed triangle, open triangle) on either BsCoaA (closed square, closed triangle) a less active protein, but it still bound the ﬂuorescent or HpCoaA (open square, open triangle) under the conditions de- ATP analog TNP-ATP with a KD similar to that of wild- scribed in (A). type, consistent with the involvement of the backbone The data are the mean values from two determinations, and the error amide, rather than the side chain, in ATP binding (Fig- bars represent the standard error. ures 5A and 5B). The SaCoaA[D6A] and SaCoaA[K13A] mutants were both catalytically inactive and failed to phosphoryl donors, was not affected by the presence or bind TNP-ATP, consistent with Asp6 coordinating the absence of the hydroxyl group at the 20 position of the Mg2+ required for the water-mediated ATP binding and ribose, and did not discriminate between purine- and with Lys13 interacting with the a- and b-phosphates of pyrimidine-based nucleotides or deoxynucleotides ATP (Figures 5A and 5B). Finally, the PaCoaA[K13A] mu- (Figure S2B). We also tried several other phosphate do- tant possessed a reduced enzymatic activity, suggest- nors, such as acetyl-phosphate, pyrophosphate, and ing a type of interaction that is not essential for ATP phosphoenolpyruvate, as well as polyphosphates that binding (Figure 5C). This is further supported by the are used by several of the microbial actin family mem- fact that this residue is not conserved among the type bers (Ding et al., 2001; Kawai et al., 2005; Tanaka et al., III CoaAs (Figure 1B). Type II and Type III Pantothenate Kinases 1255

Figure 4. The Triphosphate Binding Sites of SaCoaA and PaCoaA (A) A stereoview of the P loop region of the SaCoaAAMP-PNP complex. The Pan molecule is modeled according to the PaCoaA crystal structure (cyan carbons). Asp6 and Lys13 from the kinase domain P loop (bottom right) interact with the terminal phosphates of AMP-PNP in the usual fashion. Asn96 and Thr99 from the pseudo P loop (top left) form additional interactions with the phosphates. Glu70 is the catalytic base that activates the Pan hydroxyl. (B) A stereoview of the PaCoaA P loop region in the same orientation as the SaCoaA site shown in (A). The PaCoaA active site contains the pantothenate molecule from the PaCoaAPan complex, and the AMP- PNPMg2+ molecule is modeled according to the SaCoaA structure. The two potassium ions in sites 1 and 2 are modeled according to the Hsc70 structure (Wilbanks and McKay, 1995), and the site 2 potassium is proposed to be the monovalent cation necessary for the reaction to proceed. Note that the side chains and water molecules that interact with the modeled AMP-PNPMg2+ and the associated hydrogen bonds are not ideally modeled in the complex, and minor local structural changes will be required to opti- mally accommodate a bound nucleotide. Asp101 is spatially equivalent to Glu70 in SaCoaA, but it does not act as an essential catalytic base.

The PaCoaAATP model reveals that the interactions where the main chain and the side chain of His156 of SaCoaA and PaCoaA with the adenine ring of the have different conformations. This region is disordered bound nucleotide are significantly different (Figure 6). in monomer A of the apo PaCoaA structure. Comparison In SaCoaA, the adenine moiety is tightly bound and is of the PaCoaA and PaCoaAPan structures does not re- sandwiched between Leu11 on b2 and Tyr137 at the N veal significant changes in the dimer arrangement, but terminus of a5(Figure 6A). However, in PaCoaA, the N domains I and II are in a more ‘‘closed’’ conformation terminus of a5 is much farther away from the putative in the complex. Pantothenate binds in a buried pocket adenine binding site, and this side of the pocket opens at the dimer interface, and it is oriented such that the into a cavity lined by four arginine residues, 150, 154, C40 hydroxyl group is adjacent to the paired P loop re- 168, and 169 (Figure 6B). This open cavity does not sup- gion at the center of the molecule (Figure 7A). Secondary port tight interactions with the adenine ring, and it is structure elements that contribute to the binding pocket consistent with the high Km for ATP and the lack of nu- include strands b7 and b8, the N terminus of helix a3, the cleotide specificity (Figure S2B). In this respect, it may extended loop between b5 and a3, the N terminus and be significant that the orientation of helix a5 represents the associated loop of a60, and the C terminus and the one of the more obvious structural differences between associated loop of a40. The latter region in monomer PaCoaA and SaCoaA (Figures 2A and 2C). The Sa- A is the segment that becomes ordered in the binary CoaA[Y137A] and SaCoaA[L11A] substitutions were complex. made to verify the roles of these side chains in adenine An important feature of the complex is that both ends binding. Both mutants had compromised catalytic activ- of pantothenate are involved in specific interactions ity, but only L11A showed a significantly increased KD (Figure 7A). At one end, the C1 carboxyl group forms mul- for the environmentally sensitive fluorescence of TNP- tiple interactions with the side chains of Tyr92, Arg102, ATP (Plesniak et al., 2002; Vanoye et al., 2002)(Figures and Thr1800. Gly99 has a key role in the pocket by provid- 5A and 5B). Undetectable enhancement in the TNP- ing both space and a backbone amide nitrogen that do- ATP fluorescence in the presence of PaCoaA prevented nates an additional hydrogen bond to the C1 carboxyl analysis of this enzyme. group. At the other end, the C20 and C40 hydroxyl groups are hydrogen bonded to the side chain of Asp101, and The Pantothenate Binding Sites the C40 hydroxyl also interacts via a water molecule In the PaCoaAPan binary complex, the two monomers with the side chain of His1560 in monomer A. Asn9 is ab- are essentially identical, and the a carbons can be solutely conserved in 12 type III CoaA sequences, and it superimposed with an rmsd of 0.72 A˚ . The only signifi- is a characteristic feature of this class of enzymes (Brand cant difference occurs between residues 154 and 158, and Strauss, 2005). This residue replaces the second Structure 1256

Figure 6. The Adenine Binding Sites of SaCoaA and PaCoaA (A) SaCoaA containing the AMP-PNPMg2+ molecule from the crystal structure (gray carbons in stick representation) is shown; the interactions of the adenine ring with Tyr137 and Leu28/Leu11 are also shown. Gln125 forms a hydrogen bond with the ribose hydroxyl. (B) PaCoaA is shown with the adenine from the SaCoaA structure modeled into the active site. The adenine projects into a large cavity, created by the twisting of helix a5 relative to SaCoaA, that is lined with arginine residues. The speciﬁc hydrophobic interactions with the adenine ring that are common in kinases are not evident in PaCoaA.

glycine in the P loop DxGG(T,S)T(S,G)xxK(R,C) motif, and modeling suggests that the side chain mediates additional interactions with the g-phosphate moiety of the Figure 5. Effect of Mutations on ATP Binding and Catalysis of bound nucleotide (Figure 4B). However, our structure re- SaCoaA and PaCoaA veals that the OD1 oxygen of Asn9 also forms a key hy- (A) Binding of the ATP analog TNP-ATP to a panel of SaCoaA mu- drogen bond to the carbonyl oxygen of pantothenate tants. The identities of the individual mutant proteins are listed in (B). through a structured water molecule (Figures 4B and (B) Activity of SaCoaA mutants. Each mutant was expressed and puriﬁed, and their activities and TNP-ATP binding characteristics 7A). This structured water molecule is further anchored 0 were determined. by hydrogen bonds to the side chain of Thr157 and the (C) Activity of PaCoaA mutants. Each mutant was expressed, puri- backbone carbonyl oxygen of Val55. The poor activity ﬁed, and assayed for kinase activity. of the mutant PaCoaA[N9G] is therefore related to both The data are the mean values from two determinations, and the error Pan and phosphate binding (Figure 5C), whereas the re- bars represent the standard error. duced activity of mutant PaCoaA[T157A] is only related to Pan binding (Figure 5C). Finally, the pantothenate Type II and Type III Pantothenate Kinases 1257

Figure 7. The Pantothenate Binding Pockets in PaCoaA and SaCoaA (A) A stereoview of the PaCoaAPan binary complex showing the major hydrogen bonding and van der Waals interactions described in the text. The pocket lies at the dimer interface, and Pan is shown with cyan carbons in stick representation. Note that the carboxyl terminal end of Pan (top) is fully enclosed by the protein. (B) Modeling of pantothenate into the SaCoaA structure. The binding pocket is open to solvent, allowing the pantothenamide antimetabolites with extensions on the carboxy terminus of pantothenate to be accommodated. makes van der Waals contacts with the side chains of (<0.01% of wild-type), consistent with this residue play- Val55, Ile142, and Ile1600. ing this critical role (Figure 5B). Asp101 in PaCoaA is The structure of the SaCoaAPan complex was not structurally equivalent to Glu70 in SaCoaA (Figure 4B); obtained, but the close structural homology to PaCoaA however, the PaCoaA[D101A] mutant retained 20% of allows us to generate a model (Figure 7B). The model re- wild-type activity (Figure 5C), and this result is not con- veals that the binding pocket in SaCoaA is not enclosed sistent with a critical role for Asp101 in catalysis. This by the dimer interface, but is exposed to the solvent. observation is reminiscent of the structurally related ki- This key distinction is the result of the different mono- nase, Hsc70, which has two candidate residues for the mer-monomer arrangement in the SaCoaA dimer and catalytic base: Glu175 based on the hexokinase homol- the lack of the extended loop between b5 and a3. An im- ogy and Asp206 based on actin. Hsc70 activity is 10% of portant feature of the proposed SaCoaAPan binary the wild-type when either residue is mutated to serine or complex is that the terminal C1 carboxyl group is fully the respective amides (Wilbanks et al., 1994). The only exposed and does not interact with the protein apart other candidate for the catalytic base in PaCoaA, even from a possible salt bridge interaction with an arginine postulating a large conformational change upon ATP side chain (R113). The different architectures of the pan- binding, is the imidazole ring of His156, which forms tothenate binding pockets in PaCoaA and SaCoaA have a water-mediated hydrogen bond with the C40-OH of important implications for the catalytic mechanisms of pantothenate in monomer B. However, this interaction the two proteins and their ability to bind and process is not seen in monomer A, and the PaCoaA[H156A] mu- the pantothenamides. tant retains wild-type activity (Figure 5C). These results suggest that the type III bacterial CoaAs are mechanis- The CoaAATPPan Ternary Complex tically distinct from the type I and II enzymes and that and the Catalytic Mechanism they have active sites that more closely resemble that By structurally aligning the SaCoaAAMP-PNP and of Hsc70. PaCoaAPan complexes, it was possible to generate The structure of the Hsc70ADPPi (Wilbanks and a model of the CoaAATPPan ternary complex for McKay, 1995) shows that the nucleotide phosphates en- each enzyme. The locations of the two substrates within gage the adjacent P loops in the two kinase domains in the enzymes are shown in Figure 2, and close-ups of the the same way that AMP-PNP binds to SaCoaA. How- active site regions are shown in Figure 4. Conﬁdence in ever, Hsc70 is more similar to PaCoaA than to SaCoaA. both models derives from the close approach of the ATP Asp121 in the ‘‘pseudo P loop’’ of PaCoaA is replaced by g-phosphate to the C40 hydroxyl of pantothenate, to an asparagine (Asn96) in SaCoaA (Figure 4). An aspar- which it will be transferred in the enzyme mechanism. tate is also present at this location in Hsc70 (Asp199), Many kinases possess a Glu or Asp residue that func- where it interacts with a potassium ion that, in turn, inter- tions as a catalytic base to activate the hydroxyl of the acts with the coordination shell of the Mg2+ ion that substrate for attack on the g-phosphate of ATP (Cleland bridges the b-phosphate and the free phosphate (spa- and Hengge, 1995; Mildvan, 1997). In EcCoaA, the cata- tially equivalent to the g-phosphate) (Figure 4B). The lytic base is Asp127 (Ivey et al., 2004). In SaCoaA, muta- K+ ion is also coordinated by the side chain of a threonine genesis of Glu70 to alanine abolished enzymatic activity in Hsc70 (Thr204) that occupies the same position as the Structure 1258

C40 hydroxyl of pantothenate in the PaCoaAPan com- lized to inactive CoA analogs, and to be incorporated plex. There is no equivalent location for a monovalent into acyl carrier protein. PaCoaA was unable to phos- cation in the SaCoaAAMP-PNP complex (Figure 4A). phorylate N5-Pan in vitro, and P. aeruginosa was resis- The Hsc70[D199S] mutant displays a 100-fold reduction tant to pantothenamide growth inhibition in a broth mi- in kcat with respect to the wild-type enzyme (Wilbanks crodilution assay (Leonardi et al., 2005a; Zhang et al., et al., 1994), and the structure reveals the loss of K+ 2004). We verified that PaCoaA could be solely respon- from site 2 and a distortion of the octahedral coordina- sible for pantothenamide resistance by using the E. coli tion of Mg2+ (Johnson and McKay, 1999). Asp121 repre- strain DV70 (coaA(Ts)) mutant that is unable to grow at sents the prime candidate for the coordination of the 42C due to the inactivation of EcCoaA (Vallari and monovalent cation in PaCoaA. This residue is in a posi- Rock, 1987). Control strains were obtained by trans- tion that is predicted to modulate the interaction be- forming strain DV70 with either pPJ161, a vector harbor- tween the monovalent cation and the g-phosphate ing the E. coli coaA gene (Rock et al., 2003), or with the (Figure 4B), and the absence of activity in the Pa- empty vector, and the minimum inhibitory concentration CoaA[D121A] mutant (Figure 5C) underscores the im- for N5-Pan was determined at 42C(Figure S3A). Ex- + + portance of K /NH4 binding in catalysis. Monovalent pression of wild-type EcCoaA exacerbated the toxic cation-dependent kinases represent a distinct class of effect of the pantothenamide, consistent with an in- kinases, and crystal structures reveal that K+ and Mg2+ creased flux of antimetabolite through CoA biosynthe- interact with the triphosphate moiety of ATP to facilitate sis. Conversely, expression of PaCoaA supported the ATP binding and phosphoryl transfer (Di Cera, 2006). pantothenamide-resistant growth of strain DV70 at 42C Hsc70 is an accepted member of this subfamily, and, (Figure S3A). The presence of a type I CoaA in addition based on our structural and catalytic analyses, PaCoaA to a type III CoaA, as in B. subtilis (Brand and Strauss, is also. 2005; Yocum and Patterson, 2004), is sufficient to Enzymatic phosphoryl transfer reactions lie along a re- induce growth inhibition by the pantothenamides (data action path that is characterized as being either associa- not shown). These data demonstrated that type III tive in one extreme or dissociative in the other (Cleland CoaAs are able to confer pantothenamide resistance if and Hengge, 1995; Mildvan, 1997). Our data suggest they are the only functional pantothenate kinase. that in PaCoaA the stabilization of the metaphosphate P. aeruginosa is notoriously resistant to many antibi- leaving group by the electrostatic interactions between otics, and we therefore determined the N5-Pan sensitiv- + + 2+ the g-phosphate of ATP and the K /NH4 and Mg cat- ity of strain PAO1 transformed with pRL012, a pUCP20- ions is more important to the phosphoryl transfer than derived vector (West et al., 1994) harboring the E. coli substrate activation for nucleophilic SN2 attack by a gen- coaA gene. P. aeruginosa was resistant to N5-Pan eral base. This hypothesis leads to the conclusion that even when expressing EcCoaA capable of phosphory- PaCoaA has a more dissociative mechanism for phos- lating the antimetabolite (Figure S3B). The expression phoryl transfer compared to other kinases, like EcCoaA, of EcCoaA in the P. aeruginosa/pRL012 lysate was con- which have an associative character (Cleland and firmed by assaying the type I pantothenate kinase Hengge, 1995; Ivey et al., 2004; Mildvan, 1997). In addi- activity in the absence of monovalent cations (Fig- tion, the position of the pantothenate binding site in Pa- ure S3B, inset). These results demonstrated that P. aer- CoaA ‘‘above’’ the cleft in which ATP binds and the fact uginosa was resistant to the pantothenamides for two that the enclosed Pan site is accessible only by passing reasons: the pantothenate analogs were not taken up through the ATP site strongly suggest an ordered mech- by the pathogen, and they were not phosphorylated by anism with pantothenate binding first to the enzyme. PaCoaA. This is in contrast to EcCoaA, which also operates via Our structure of the PaCoaAPan complex provides an ordered mechanism, but with ATP as the leading an explanation for the resistance of this enzyme to pan- substrate (Song and Jackowski, 1994). In the case of tothenate analogs. The pantothenamides have substitu- SaCoaA, the different dimer interface creates two sol- ents that modify the carboxyl (C1) end of pantothenate, vent-exposed openings to the active site, and it is and the tight binding pocket cannot accommodate likely that ATP enters from one direction and Pan binds these substituents. In contrast, this end of the through the other opening in a nonsequential mecha- SaCoaAPan binding pocket is fully exposed and able nism. Finally, kinases typically display a conformational to bind the extended analogs. closing of the active site cleft to exclude water during phosphoryl transfer. Although this is likely to occur Insights into Human PanK2 Structure and Function upon formation of the ternary complex for which we do The inherited PKAN neurodegenerative disorder is linked not have a structure, it is relevant that the Pan bound to mutations in the human PANK2 gene. All four isoforms form of PaCoaA is in a more closed conformation, and of human pantothenate kinases have a common cata- a similar interdomain movement has recently been re- lytic core domain, which shows >80% sequence identity ported for Hsc70 (Jiang et al., 2005). between PanK1a, PanK1b, PanK2, and PanK3 (Ho¨ rtnagel et al., 2003; Rock et al., 2002; Zhang et al., 2005), Role of PaCoaA in the P. aeruginosa and they also have amino acid sequences inserted be- Resistance to N5-Pan tween strands b2 and b3 and between helices a1 and Pantothenamides inhibit E. coli and S. aureus growth by a2ofSaCoaA (Figure 1A). The eukaryotic PanKs differ blocking fatty acid biosynthesis (Leonardi et al., 2005a; from SaCoaA in that they are subject to feedback inhibi- Zhang et al., 2004). Pantothenamide phosphorylation tion by CoA and its thioesters (Leonardi et al., 2005b), by the type I and type II CoaAs allows these compounds and these inserted sequences may correspond to the to enter the CoA biosynthetic pathway, to be metabo- structural elements required for acetyl-CoA regulation Type II and Type III Pantothenate Kinases 1259

PanK2[L563P] mutant is catalytically inactive (Figure S4), and the SaCoaA[L263P] mutant is catalytically compromised (Figure 5B). In the SaCoaA structure, the side chain of Leu263 interacts with the backbone of strand b1(Figure 8), which contains the Mg2+-coordinating Asp6 and the phosphate binding Lys13 (Figure 4A). The substitution of Leu263 with proline is predicted to alter the orientation of strand b1 and directly affect the structure of the ATP binding pocket. Indeed, the SaCoaA[L263P] mutant has a considerably higher KD for TNP-ATP (Figure 5A), accounting for its lower catalytic activity.

Experimental Procedures

Materials and Plasmid Construction Materials and plasmid construction are described in the Supple- mental Data.

Protein Expression and Purification Protein expression and purification are described in the Supplemen- tal Data. Figure 8. PKAN-Linked Mutations Mapped onto the SaCoaA Structure Crystallization Cocrystals of the Se-Met SaCoaA with AMP-PNP and Mg2+ were In the SaCoaA structure, AMP-PNP is represented with gray carbons grown by the sitting drop vapor diffusion method. Prior to crystalliza- and the Mg2+ is a green sphere. Residues Gly8 and Gly224 corre- tion, the protein was incubated overnight with 30 mM AMP-PNP and spond to residues Gly219 and Gly521 in human PanK2 and are lo- 30 mM MgCl . Typically, 1 ml precipitant solution was added to 1 ml cated in loops that bracket the ATP binding site. These glycines are 2 protein, and the mixture was allowed to equilibrate over 1 ml precip- important to the structure of the ATP binding fold, and the absence itant solution of 28% (w/v) PEG 2000 MME and 0.1 M bis-Tris (pH 6.5), of a side chain allows room for the nucleotide to bind. Leu263, at 18C. After several days, hexagonal crystals appeared and were located in helix a8, corresponds to Leu563 in PanK2 and interacts allowed to grow for 1 week. All crystals were cryoprotected in the with strand b1 to position the conserved DxGG(T,S)T(S,G)xxK(R,C) 50/50 mixture of Paratone-N and mineral oil and were flash frozen motif to bind the phosphates of the nucleotide. in liquid nitrogen. Crystals of Se-Met PaCoaA were grown by using the hanging- drop vapor diffusion technique. Hanging drops were prepared by of enzyme activity (Zhang et al., 2005). The SaCoaA point mixing 4 ml protein solution with 1 ml water and were allowed to equil- mutants [K13A] and [E70A] are equivalent to the PanK2 ibrate over water at 18C. A crystal of the Pan bound Se-Met enzyme point mutants [K224A] and [E338A], and these PanK2 was obtained by soaking the PaCoaA crystal in 40 mM Tris-HCl (pH mutants are inactive (Figure S4). Similarly, mutation of 7.5), 1 mM DTT, 1 mM EDTA, and 10 mM pantothenate for 1 day at the PanK2 Leu222 to Ala, equivalent to Leu11 in SaCoaA, 18C. All crystals were cryoprotected in 25 mM Tris-HCl (pH 7.5), did not inactivate the enzyme. This supports both the 0.5 mM DTT, 26% ethylene glycol, and 26% glycerol and were flash frozen in liquid nitrogen. structural and functional conclusions that SaCoaA and PanK2 are closely related. There are three amino acids Data Collection and Structure Determination that are identical in SaCoaA and PanK2 and that are mu- of the SaCoaAAMP-PNP Complex tated in PKAN patients (Figure 1A). The most common Data were collected on the Advanced Photon Source (beamline 19- mutation is PanK2[G521R], which is catalytically inactive BM) at the Argonne National Laboratory. Two data sets were col- and fails to fold properly (Zhang et al., 2006). Gly521 cor- lected at 100K for the Se-Met protein crystal, and one data set responds to Gly224 of SaCoaA, located in the loop be- was collected for a native crystal. Diffraction data were integrated, tween strand b9 and 3/10 helix a8(Figure 8), and it has reduced, and scaled with DENZO and SCALEPACK (Otwinowski and Minor, 1997). The crystals belonged to space group P63 with both functional and structural roles. Functionally, the ab- unit cell dimensions of a = 104.0 A˚ and c = 85.7 A˚ . We used the pro- sence of a side chain on Gly224 provides space for the gram SOLVE (Terwilliger and Berendzen, 1999) to scale two wave- 2+ binding of Mg and the phosphate groups of ATP. Struc- length data sets together, to locate selenium positions, and to calcu- turally, Gly224 is located at the N terminus of helix a8 and late the protein phase angles. The model was completed in several may be important in folding. This conclusion is consis- repetitions of manual building with O (Jones et al., 1991) and crystal- tent with the observation that SaCoaA[G224R] is insolu- lographic refinement with REFMAC5 (Murshudov et al., 1999). The fi- nal refinement statistics are shown in Table S1. Accessible and bur- ble, reflecting the properties of PanK2[G521R]. A less ied surface areas were calculated by using the program AREAIMOL common PKAN mutation is PanK2[G219V], which is (CCP4, 1994) with a probe radius of 1.4 A˚ . also catalytically inactive (Zhang et al., 2006). Gly219 of human PanK2 corresponds to Gly8 of SaCoaA, located Data Collection and Structure Determination in the P loop between strands b1 and b2(Figure 8). This of the Apo and Pan Bound PaCoaA residue creates space for the binding of the ATP phos- MAD data sets were collected at the Southeast Regional Collabora- phates, and the introduction of a side chain would inter- tive Access Team (SER-CAT) 22-BM beamline at the Advanced fere with ATP binding. SaCoaA[G8V] did not fold properly Photon Source, Argonne National Laboratory. Three wavelengths 0.9795 A˚ (the peak), 0.9797 A˚ (the inflection), and 0.9718 A˚ (the and was recovered as an insoluble precipitate. Finally, remote) were selected based on the fluorescence scan of the Se K Leu563 in human PanK2 corresponds to Leu263 of absorption edge. The higher-resolution data for PaCoaA and SaCoaA, located at the end of the C-terminal helix. The PaCoaAPan were collected at the (SER-CAT) 22-ID beamline. All Structure 1260

data sets were collected at 100K and were processed by using the Brunger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., HKL2000 program package (Otwinowski and Minor, 1997). Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., The program SOLVE (Terwilliger and Berendzen, 1999) was used Pannu, N.S., et al. (1998). Crystallography & NMR system: a new to locate the 12 selenium sites and to calculate initial phases in the software suite for macromolecular structure determination. Acta resolution range of 50–2.5 A˚ . The program RESOLVE (Terwilliger, Crystallogr. D Biol. Crystallogr. 54, 905–921. 2000) was used for density modification, and all of the ordered res- Calder, R.B., Williams, R.S.B., Ramaswamy, G., Rock, C.O., Camp- idues could be built into the resulting electron density map by using bell, E., Unkles, S.E., Kinghorn, J.R., and Jackowski, S. (1999). Clon- ˚ Xtalview (McRee, 1999). The PaCoaA structure was refined to 2.2 A ing and characterization of a eukaryotic pantothenate kinase gene resolution with the higher-resolution data sets by using the program (panK) from Aspergillus nidulans. J. Biol. Chem. 274, 2014–2020. CNS (Brunger et al., 1998). The crystal asymmetric unit contained one copy of the biological dimer. Pertinent data collection, phasing, CCP4 (Collaborative Computation Project, Number 4) (1994). The and refinement statistics are shown in Table S2. The PaCoaA and CCP4 suite: programs for protein crystallography. Acta Crystallogr. PaCoaAPan crystals were isomorphous, and the PaCoaAPan D Biol. Crystallogr. 50, 760–763. structure was determined by direct refinement of the dimeric struc- Cheek, S., Zhang, H., and Grishin, N.V. (2002). Sequence and structure of PaCoaA against the 1.9 A˚ diffraction data obtained for the ture classification of kinases. J. Mol. Biol. 320, 855–881. PaCoaAPan binary complex. After the first round of refinement, Cheek, S., Ginalski, K., Zhang, H., and Grishin, N.V. (2005). A com- the Fo 2 Fc difference map clearly revealed strong positive density prehensive update of the sequence and structure classification of for pantothenate and also showed residues 155–163, which were kinases. BMC Struct. Biol. 5,6. not visible in one of the subunits of the apo PaCoaA structure. Pertinent statistics are also shown in Table S2. The program Cleland, W.W., and Hengge, A.C. (1995). Mechanisms of phosphoryl PROCHECK (Laskowski et al., 1993) was used to evaluate the quality and acyl transfer. FASEB J. 9, 1585–1594. of the structures. All of the figures were produced by using delCardayre, S.B., Stock, K.P., Newton, G.L., Fahey, R.C., and Da- MOLSCRIPT (Kraulis, 1991) and were rendered with RASTER3D vies, J.E. (1998). Coenzyme A disulfide reductase, the primary low (Merritt and Bacon, 1997). molecular weight disulfide reductase from Staphylococcus aureus. Purification and characterization of the native enzyme. J. Biol. Analytical Gel Filtration and Ultracentrifugation Chem. 273, 5744–5751. Analytical gel filtration and ultracentrifugation are described in the Di Cera, E. (2006). Enzymes activated by monovalent cations: Supplemental Data. a structural perspective. J. Biol. Chem. 281, 1305–1308. Ding, Y.R., Ronimus, R.S., and Morgan, H.W. (2001). Thermotoga Kinetics, Enzymatic, and ATP Binding Assays maritima phosphofructokinases: expression and characterization Kinetics, enzymatic, and ATP binding assays are described in the of two unique enzymes. J. Bacteriol. 183, 791–794. Supplemental Data. Ho¨ rtnagel, K., Prokisch, H., and Meitinger, T. (2003). An isoform of hPANK2, deficient in pantothenate kinase-associated neurodegen- Expression of the Human PanK2 Mutants eration, localizes to mitochondria. Hum. Mol. Genet. 12, 321–327. Expression of the human PanK2 mutants is described in the Supple- mental Data. Ivey, R.A., Zhang, Y.-M., Virga, K.G., Hevener, K., Lee, R.E., Rock, C.O., Jackowski, S., and Park, H.-W. (2004). The structure of the Supplemental Data pantothenate kinase-ADP-pantothenate ternary complex reveals Supplemental Data include Supplemental Experimental Procedures, the relationship between the binding sites for substrate, allosteric two crystallographic tables, and four additional figures and are regulator and antimetabolites. J. Biol. Chem. 279, 35622–35629. available at http://www.structure.org/cgi/content/full/14/8/1251/ Jiang, J., Prasad, K., Lafer, E.M., and Sousa, R. (2005). Structural ba- DC1/. sis of interdomain communication in the Hsc70 chaperone. Mol. Cell 20, 513–524. Acknowledgments Johnson, E.R., and McKay, D.B. (1999). Mapping the role of active site residues for transducing an ATP-induced conformational We thank Karen Miller and Ruobing Zhou for their expert technical change in the bovine 70-kDa heat shock cognate protein. Biochem- assistance, Dr. Amanda Nourse for the ultracentrifugation data, istry 38, 10823–10830. and Dr. Changsoo Chang at the Structural Biology Center (SBC) Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). Im- beamline 19-BM at the Argonne National Laboratory for help with proved methods for binding protein models in electron density data collection. Use of the SBC 19-BM and Southeast Regional Col- maps and the location of errors in these models. Acta Crystallogr. laborative Access Team (SER-CAT) beamlines at the Advanced A 47, 110–119. Photon Source was supported by the U.S. Department of Energy, Basic Energy Sciences, Office of Science, under contract no. Kabsch, W., and Holmes, K.C. (1995). The actin fold. FASEB J. 9, W-31-109-Eng-38. SER-CAT-supporting institutions may be found 167–174. at www.ser.anl.gov/new/index.html. The authors are supported by Kachmer, J.F., and Boyer, P.D. (1953). Kinetic analysis of enzyme National Institutes of Health grants GM 62896 (S.J.) and GM 34496 reactions. II. The potassium activation and calcium inhibition of (C.O.R.), Cancer Center (CORE) Support grant CA 21765, and the pyruvic phosphoferase. J. Biol. Chem. 200, 669–682. American Lebanese Syrian Associated Charities. Kawai, S., Mukai, T., Mori, S., Mikami, B., and Murata, K. (2005). Hypothesis: structures, evolution, and ancestor of glucose kinases Received: March 31, 2006 in the hexokinase family. J. Biosci. Bioeng. 99, 320–330. Revised: June 6, 2006 Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both Accepted: June 9, 2006 detailed and schematic plots of protein structures. J. Appl. Crystal- Published: August 15, 2006 logr. 24, 946–950. Laskowski, R.A., McArthur, M.W., Moss, D.S., and Thornton, J.M. References (1993). PROCHECK: a program to check the quality of protein structures. J. Appl. Crystallogr. 26, 282–291. Bork, P., Sander, C., and Valencia, A. (1992). An ATPase domain common to prokaryotic cell cycle proteins, sugar kinases, actin, Leonardi, R., Chohnan, S., Zhang, Y.-M., Virga, K.G., Lee, R.E., Rock, and hsp70 heat shock proteins. Proc. Natl. Acad. Sci. USA 89, C.O., and Jackowski, S. (2005a). A pantothenate kinase from Staph- 7290–7294. ylococcus aureus refractory to feedback regulation by coenzyme A. Brand, L.A., and Strauss, E. (2005). Characterization of a new panto- J. Biol. Chem. 280, 3314–3322. thenate kinase isoform from Helicobacter pylori. J. Biol. Chem. 280, Leonardi, R., Zhang, Y.-M., Rock, C.O., and Jackowski, S. (2005b). 20185–20188. Coenzyme A: back in action. Prog. Lipid Res. 44, 125–153. Type II and Type III Pantothenate Kinases 1261

McRee, D.E. (1999). XtalView/Xfit—a versatile program for manipu- Wilbanks, S.M., Luca-Flaherty, C., and McKay, D.B. (1994). Struc- lating atomic coordinates and electron density. J. Struct. Biol. 125, tural basis of the 70-kilodalton heat shock cognate protein ATP 156–165. hydrolytic activity. I. Kinetic analyses of active site mutants. J. Biol. Merritt, E.A., and Bacon, D.J. (1997). Raster3D: photorealistic mo- Chem. 269, 12893–12898. lecular graphics. Methods Enzymol. 277, 505–524. Yocum, R.R., and Patterson, T.A. November 2004. U.S. patent 6,830,898. Mildvan, A.S. (1997). Mechanisms of signaling and related enzymes. Proteins 29, 401–416. Yun, M., Park, C.-G., Kim, J.-Y., Rock, C.O., Jackowski, S., and Park, H.-W. (2000). Structural basis for the feedback regulation of Escher- Murshudov, G.N., Vagin, A.A., Lebedev, A., Wilson, K.S., and Dod- icia coli pantothenate kinase by coenzyme A. J. Biol. Chem. 275, son, E.J. (1999). Efficient anisotropic refinement of macromolecular 28093–28099. structures using FFT. Acta Crystallogr. D Biol. Crystallogr. 55, 247– 255. Zhang, Y.-M., Frank, M.W., Virga, K.G., Lee, R.E., Rock, C.O., and Jackowski, S. (2004). Acyl carrier protein is a cellular target for the Oria-Hernandez, J., Cabrera, N., Perez-Montfort, R., and Ramirez- antibacterial action of the pantothenamide class of pantothenate an- Silva, L. (2005). Pyruvate kinase revisited: the activating effect of timetabolites. J. Biol. Chem. 279, 50969–50975. K+. J. Biol. Chem. 280, 37924–37929. Zhang, Y.-M., Rock, C.O., and Jackowski, S. (2005). Feedback reg- Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction ulation of murine pantothenate kinase 3 by coenzyme A and coen- data collected in oscillation mode. Methods Enzymol. 276, 307–326. zyme A thioesters. J. Biol. Chem. 280, 32594–32601. Plesniak, L., Horiuchi, Y., Sem, D., Meinenger, D., Stiles, L., Shaffer, Zhang, Y.-M., Rock, C.O., and Jackowski, S. (2006). Biochemical J., Jennings, P.A., and Adams, J.A. (2002). Probing the nucleotide properties of human pantothenate kinase 2 isoforms and mutations binding domain of the osmoregulator EnvZ using fluorescent nucle- linked to pantothenate kinase-associated neurodegeneration. otide derivatives. Biochemistry 41, 13876–13882. J. Biol. Chem. 281, 107–114. Rock, C.O., Calder, R.B., Karim, M.A., and Jackowski, S. (2000). Zhou, B., Westaway, S.K., Levinson, B., Johnson, M.A., Gitschier, J., Pantothenate kinase regulation of the intracellular concentration of and Hayflick, S.J. (2001). A novel pantothenate kinase gene (PANK2) coenzyme A. J. Biol. Chem. 275, 1377–1383. is defective in Hallervorden-Spatz syndrome. Nat. Genet. 28, 345– Rock, C.O., Karim, M.A., Zhang, Y.-M., and Jackowski, S. (2002). The 349. murine Pank1 gene encodes two differentially regulated pantothenate kinase isozymes. Gene 291, 35–43. Accession Numbers Rock, C.O., Park, H.-W., and Jackowski, S. (2003). Role of feedback regulation of pantothenate kinase (CoaA) in the control of coenzyme The structure factors and refined coordinates have been deposited A levels in Escherichia coli. J. Bacteriol. 185, 3410–3415. in the PDB with the accession codes 2EWS, 2F9T, and 2F9W for the Song, W.-J., and Jackowski, S. (1992). Cloning, sequencing, and SaCoAAMP-PNP complex, PaCoaA, and the PaCoaAPan com- expression of the pantothenate kinase (coaA) gene of Escherichia plex, respectively. coli. J. Bacteriol. 174, 6411–6417. Song, W.-J., and Jackowski, S. (1994). Kinetics and regulation of pantothenate kinase from Escherichia coli. J. Biol. Chem. 269, 27051–27058. Strauss, E., and Begley, T.P. (2002). The antibiotic activity of N-pen- tylpantothenamide results from its conversion to ethyldethia-coenzyme A, a coenzyme A antimetabolite. J. Biol. Chem. 277, 48205– 48209. Suelter, C.H. (1974). Monovalent cations in enzyme-catalyzed reactions. In High Molecular Complexes, H. Sigel, ed. (New York: Marcel Dekker, Inc.), pp. 201–251. Tanaka, S., Lee, S.O., Hamaoka, K., Kato, J., Takiguchi, N., Naka- mura, K., Ohtake, H., and Kuroda, A. (2003). Strictly polyphosphate- dependent glucokinase in a polyphosphate-accumulating bacterium, Microlunatus phosphovorus. J. Bacteriol. 185, 5654–5656. Terwilliger, T.C. (2000). Maximum-likelihood density modification. Acta Crystallogr. D Biol. Crystallogr. 56, 965–972. Terwilliger, T., and Berendzen, J. (1999). Automated MAD and MIR structure solution. Acta Crystallogr. D Biol. Crystallogr. 55, 501–505. Vallari, D.S., and Rock, C.O. (1987). Isolation and characterization of temperature-sensitive pantothenate kinase (coaA) mutants of Escherichia coli. J. Bacteriol. 169, 5795–5800. Vallari, D.S., Jackowski, S., and Rock, C.O. (1987). Regulation of pantothenate kinase by coenzyme A and its thioesters. J. Biol. Chem. 262, 2468–2471. Vanoye, C.G., MacGregor, G.G., Dong, K., Tang, L., Buschmann, A.S., Hall, A.E., Lu, M., Giebisch, G., and Hebert, S.C. (2002). The carboxyl termini of K(ATP) channels bind nucleotides. J. Biol. Chem. 277, 23260–23270. West, S.E., Schweizer, H.P., Dall, C., Sample, A.K., and Runyen-Ja- necky, L.J. (1994). Construction of improved Escherichia-Pseudo- monas shuttle vectors derived from pUC18/19 and sequence of the region required for their replication in Pseudomonas aeruginosa. Gene 148, 81–86. Wilbanks, S.M., and McKay, D.B. (1995). How potassium affects the activity of the molecular chaperone Hsc70. II. Potassium binds spe- cifically in the ATPase active site. J. Biol. Chem. 270, 2251–2257.