Structure and Mechanism of a Eukaryotic FMN Adenylyltransferase

doi:10.1016/j.jmb.2009.04.022 J. Mol. Biol. (2009) 389, 388–400

Available online at www.sciencedirect.com

Structure and Mechanism of a EukaryoticFMN Adenylyltransferase

Carlos Huerta1, Dominika Borek1, Mischa Machius1,Nick V. Grishin1,2 and Hong Zhang1⁎
1Department of Biochemistry,University of TexasSouthwestern Medical Center,Dallas, TX 75390, USA2Howard Hughes MedicalInstitute, University of TexasSouthwestern Medical Center,Dallas, TX 75390, USA
Received 11 February 2009;received in revised form8 April 2009;accepted 11 April 2009Available online16 April 2009

*Corresponding author. E-mail [email protected] used: FMNAT, flav

adenylyltransferase; FAD, flavin adeFMN, flavin mononucleotide; RFK,FADS, flavin adenine dinucleotide s3′-phosphoadenosine 5′-phosphosupyrophosphate; CgFMNAT, CandidaAMPCPP, α,β-methyleneadenosinesingle-wavelength anomalous dispe5′-phosphosulfate; PDB, Protein Danicotinamide mononucleotide adenyTmFADS, Thermotoga maritima FADS

0022-2836/$ - see front matter © 2009 E

Flavin mononucleotide adenylyltransferase (FMNAT) catalyzes the forma-tion of the essential flavocoenzyme flavin adenine dinucleotide (FAD) andplays an important role in flavocoenzyme homeostasis regulation. Bysequence comparison, bacterial and eukaryotic FMNAT enzymes belong totwo different protein superfamilies and apparently utilize different sets ofactive-site residues to accomplish the same chemistry. Here we report thefirst structural characterization of a eukaryotic FMNAT from the pathogenicyeast Candida glabrata. Four crystal structures of C. glabrata FMNAT indifferent complexed forms were determined at 1.20–1.95 Å resolutions,capturing the enzyme active-site states prior to and after catalysis. Thesestructures reveal a novel flavin-binding mode and a unique enzyme-boundFAD conformation. Comparison of the bacterial and eukaryotic FMNATsprovides a structural basis for understanding the convergent evolution ofthe same FMNAT activity from different protein ancestors. Structure-basedinvestigation of the kinetic properties of FMNAT should offer insights intothe regulatory mechanisms of FAD homeostasis by FMNAT in eukaryoticorganisms.

© 2009 Elsevier Ltd. All rights reserved.

Keywords: flavocoenzymes; FAD biosynthesis; adenylyltransferase;Rossmann-like fold; convergent evolution
Edited by G. Schulz
Introduction

Flavin mononucleotide (FMN) and flavin adeninedinucleotide (FAD) are essential cofactors that areinvolved in many redox reactions in the cell.1 Thechemical and functional versatility of these cofac-tors, in association with various flavoproteins,allows them to be involved in a large variety ofreaction types and to participate in many cellular

ess:

in mononucleotidenine dinucleotide;riboflavin kinase;ynthetase; PAPS,lfate; PPi, inorganicglabrata FMNAT;5′-triphosphate; SAD,rsion; APS, adenosineta Bank; NMNAT,lyltransferase;.

lsevier Ltd. All rights reserve

processes ranging from energy production, lightemission, DNA repair, chromatin remodeling, andprotein folding to detoxification, neural deve-lopment, and apoptosis.2–4 Riboflavin, also knownas vitamin B2, is the universal precursor for thesynthesis of FMN and FAD, the primary form offlavins in cells.5,6 Due to the involvement of flavo-cofactors in wide-ranging metabolic processes,riboflavin deficiency leads to a multitude of physio-logical aberrations such as abnormal fetal develop-ment, inadequate ion absorption, cardiovasculardisease, and corneal defects.7In prokaryotes, yeast, and plants, riboflavin is

either synthesized de novo or obtained from theenvironment and transported into cells.6,8 Highereukaryotes, such as humans, lack the de novo ribo-flavin synthesis machinery, and their only means ofobtaining riboflavin is through the diet.5,6 The ribo-flavin transporter for eukaryotes has been identifiedin Saccharomyces cerevisiae and is encoded by thegene mch5.9 Converting riboflavin into FAD in-volves two universally conserved enzymes: ribofla-vin kinase (RFK) and flavin mononucleotideadenylyltransferase (FMNAT) (Fig. 1). RFK (ATP:riboflavin 5′-phosphotransferase; EC 2.7.1.26) phos-

d.

Fig. 1. Reactions catalyzed by RFK and FMNAT.

389Structure of Yeast FMN Adenylyltransferase

phorylates riboflavin to generate FMN, whileFMNAT (ATP:FMN adenylyltransferase; EC 2.7.7.2)adenylates FMN to form FAD. In bacteria, RFK andFMNAT are encoded in the same gene (ribF orribC),10,11 and the protein product is referred to asflavin adenine dinucleotide synthetase (FADS), withthe FMNAT domain located at the N-terminus andthe RFK domain at the C-terminus. In eukaryotes,RFK and FMNATare encoded by separate genes.12,13

In higher eukaryotes, the gene encoding FMNATalsocontains a second domain with sequence similarityto proteins involved in molybdenum cofactor bio-synthesis such as MogA and MoeA.14,15 The poten-tial function of this molybdenum-cofactor-bindingprotein-like domain is unknown.Bacterial and eukaryotic RFKs are similar in

sequence and structure, and they belong to aunique protein family containing only RFKs.16,17In contrast, the evolutionary link between bacterialand eukaryotic FMNATs is less clear, as they showlittle sequence similarity and are classified in diffe-rent protein superfamilies in SCOP18 or in differentclans in Pfam19 databases. The bacterial FMNATdomain of the bifunctional RFK/FMNAT belongs tothe (H/T)xGH motif-containing nucleotidylyl trans-ferase superfamily, while eukaryotic FMNAT iscurrently classified as a member of the 3′-phospho-adenosine 5′-phosphosulfate (PAPS) reductase-likefamily belonging to the “adenine nucleotide α-

hydrolase-like” superfamily, which has conservedmotifs different from those of nucleotidylyl trans-ferases. Despite substantial differences in sequenceand structure, mammalian and bacterial FMNATenzymes have similar kinetic properties.20,21 Bothenzymes catalyze the formation of FAD through anordered bi–bi mechanism and have the samesubstrate binding and product release order,where ATP binds first to the enzyme followedby FMN, and product inorganic pyrophosphate(PPi) is released first followed by FAD release.Pronounced product feedback inhibition wasobserved for rat liver FMNAT, and it wassuggested that such a property would enableFMNAT to play a role in regulating cellular FADhomeostasis as the Ki values of FAD againstFMN (0.75 μM) and Mg2+-ATP (1.3 μM) are closeto the concentration of free FAD (0.4 μM).21 Main-tenance of FAD homeostasis is important, as severalcellular processes, such as oxidative protein foldingand homocysteine metabolism, are sensitive to FADlevels.22,23Both RFK and FMNAT are needed to generate

the indispensable flavocofactors FMN and FAD.6

The essentiality of the two enzymes has been esta-blished experimentally in bacterial24 and yeast spe-cies,12,13,22 and has been inferred to all otherorganisms. The significant differences betweeneukaryotic and bacterial FMNATs make them

390 Structure of Yeast FMN Adenylyltransferase

particularly attractive targets for developing selec-tive anti-infectious drugs.24 Structural analysis ofboth eukaryotic and bacterial FMNATs will revealthe different configurations of the substrate-bindingand catalytic sites, which may benefit structure-based inhibitor development efforts. Such analysiswill also address two fundamental questions: howeukaryotic and bacterial FMNAT accomplish thesame chemistry with different active-site architec-tures, and what mechanistic controls are embeddedin the enzyme to influence FAD homeostasis. Herewe report the first structural characterization of aeukaryotic FMNAT from Candida glabrata, anopportunistic yeast pathogen that causes candide-mia and invasive candidiasis.25,26 We have deter-mined the crystal structures of C. glabrata FMNAT(CgFMNAT) in the apo form and in three differentcomplexed forms [with ATP, with substrate FMNand the ATP analog α,β-methyleneadenosine 5′-triphosphate (AMPCPP), and with products FADand PPi, respectively]. These structures reveal anovel flavin-binding mode and a detailed catalyticsite configuration that are likely shared among alleukaryotic FMNATs. Combined with results from asteady-state kinetic analysis, a mechanism for eu-

Table 1. Data collection and refinement statistics

ProteinSelenomethionylapo-CgFMNAT Apo-CgFMNAT

Space group/monomerper asymmetric unit

P3221/1 P3221/1

Data SAD NativeWavelength (Å) 0.97927 0.97931Resolution (Å) 50.0–2.18 50.0–1.20Unit cell dimensions

a, b, c (Å) 80.31, 80.31, 78.27 80.09, 80.09, 78.09α, β, γ (°) 90.0, 90.0, 120.0 90.0, 90.0, 120.0

Total number of reflections 200,196 1,036,966Number of unique reflections 15,582 90,728% Completeness 99.9 (100.0)a 99.9 (99.7)Rsym

b 0.091 (0.335) 0.070 (0.442)I/σ 38.9 (8.5) 48.2 (2.3)Mosaicity (°) 0.6 0.4Model refinement Anisotropicc

Resolution range (Å) 50.0–1.20Rwork

d (%) 16.2Rfree

e (%) 17.8Number of protein atoms 2502Number of ligand atoms 13Number of water molecules 390Average B-factor (Å2)

Protein atoms 17.1Ligand atoms 21.3Water molecules 29.3

rmsd bond length (Å) 0.015rmsd bond angle (°) 1.621Ramachandran plot

Favored region (%) 98.6Allowed region (%) 1.4Outliers (%) 0.0a Values in parenthesis are for the highest-resolution shell.b Rsym=∑hkl[(∑j(|Ij− ⟨I⟩|)/∑jIj ], where ⟨I⟩ is the average of all j mc Six anisotropic thermal factors were used for selected atoms, while

thermal parameters were evaluated and assigned using PARVATI27 and Rwork=∑hkl|Fo−kFc|/∑hkl|Fo|, where Fo and Fc are the observede R-factor calculated from randomly selected 1.5% (apo-CgFMNAT

for cross-validation.

karyotic FMNAT-catalyzed adenylyl transfer reac-tion is proposed.

Results

Quality of structures

The crystal structure of apo-CgFMNATwas solvedby the single-wavelength anomalous dispersion(SAD) phasing method using selenomethionine asthe source of anomalous dispersion and was refinedagainst a native data set to a resolution of 1.20 Å(Table 1). There is one CgFMNAT molecule in theasymmetric unit. Crystal-packing analysis suggeststhat the functional unit of CgFMNAT is a monomer,consistent with the gel-filtration result showing thatCgFMNAT is monomeric in solution (see Fig. S1,available online). The refined model contains resi-dues −3 to 304, except for residues 85–101, for whichwe found no associated electron density and whichare presumably disordered. Residues −3 to 0 (Gly-Ala-Met-Val) were introduced during cloning (seeSupplementary Methods). The CgFMNAT–ATPcomplex crystal is isomorphous to apo-CgFMNAT,

ATPcomplex

Ternary complexwith substrates

Ternary complexwith products

P3221/1 C2/6 C2/6

Native Native Native1.54178 1.54178 0.9787450.0–1.87 50.0–1.95 50.0–1.35

79.79, 79.79, 77.94 207.83, 81.75, 136.70 206.58, 81.48, 136.6090.0, 90.0, 120.0 90.0, 129.67, 90.0 90.0, 129.79, 90.0

124,769 312,183 2,800,90624,161 123,738 380,943

99.9 (99.6) 96.4 (97.1) 100 (99.8)0.078 (0.549) 0.088 (0.447) 0.077 (0.551)22.0 (2.5) 11.5 (2.1) 35.9 (2.4)

0.7 0.5 0.4Isotropic Isotropic Anisotropic50.0–1.87 35.1–1.95 32.2–1.35

17.4 17.5 15.322.2 23.6 18.72397 14,700 14,96844 397 417303 1790 1870

19.3 18.2 17.123.5 17.5 15.830.8 28.5 30.40.015 0.014 0.0141.536 1.458 1.523

98.6 98.9 99.11.05 1.1 0.90.35 0.0 0.0

easurements of reflection hkl.isotropic B-factors were used for the remaining atoms. Anisotropicd ANISOANL.28

and calculated structure factors, respectively.) or 5% (complexes) reflections that are excluded from refinement

Fig. 2. Electron densities of bound ligands and comparison of CgFMNAT structures. Simulated-annealing Fo−Fc omitmaps of (a) ATP in the binary complex; (b) AMPCPP and FMN in the substrate ternary complex; and (c) FAD and PPi inthe product ternary complex. The maps are contoured at 3.0σ (a and c) and 2.5σ (b). (d) Superposition of apo form(magenta), ATP complex (tan), substrate ternary complex monomer C (dark yellow), and product ternary complexmonomer B (dark blue).


and its model was refined to 1.87 Å resolution. Thismodel contains residues −3 to 83, residues 103–304,and an ATP molecule with well-defined density(Fig. 2a). Crystals of the substrate and productternary complexes are isomorphous to each otherand belong to the space group C2, with six mono-mers in the asymmetric unit. They were refined toresolutions of 1.95 Å and 1.35 Å, respectively. Forthe substrate ternary complex, the densities forAMPCPP and Mg2+ are well defined, and double

Fig. 3. Overall structure of CgFMNAT. (a) Ribbon diagramproducts FAD, PPi, and Mg2+. Secondary structure elements aand catalysis are labeled and highlighted in different colors, wmotif in magenta, γ-phosphate motif in red, and flavin motif i(b) Comparison of CgFMNAT with closely related APS reducremoved for clarity. The bound ligand FAD in CgFMNAT anatom type, with carbon atoms shown in yellow and orangsimilarly. The characteristic PP-loop motifs are shown in retopology are shown in magenta.

conformations of the AMPCPP phosphate tail areobserved (Fig. 2b). The density for the phospho-ribityl tail of FMN, on the other hand, is discon-tinuous, indicating significant conformationalflexibility (Fig. 2b). For the product ternary com-plex, the densities for both products FAD andpyrophosphate (PPi) are well defined (Fig. 2c). Theflexible loop region (residues 84–100) disordered inthe apo and ATP-complexed CgFMNAT structuresis ordered in two of the six crystallographically

of the CgFMNAT structure shown in complex with there labeled. Structure motifs involved in substrate bindingith PP-loop in dark blue, ADE motif in dark green, ARG1n brown. The magnesium ion is shown as a green sphere.tase. Helix α1 and the last ∼70 residues of CgFMNAT ared APS in APS reductase are shown as sticks colored bye, respectively. Equivalent structural motifs are coloredd. Regions that deviate from the typical Rossmann-fold


independent CgFMNAT monomers in the ternarycomplexes, presumably due to crystal-packinginteractions. For all models, the main-chain dihe-dral angles (ϕ,ψ) for each residue are within theallowed regions of Ramachandran plot. The excep-tion is Gly224 in the ATP complex structure, whichis associated with well-defined electron density.Superposition of the Cα backbones of the apoprotein, ATP complex, and substrate ternary com-plex (monomer C) to the product ternary complex(monomer B) gives root mean square deviation(rmsd) values of 0.42 Å, 0.41 Å, and 0.27 Å, res-pectively, indicating no substantial conformationalchanges among these structures (Fig. 2d).

Overall structure

CgFMNAT is composed of two domains (Fig. 3a).The N-terminal domain has an α/β fold with a

Fig. 4. Substrate and product binding in CgFMNAT. (a) Decatalytically relevant conformer I is shown. Protein residues inis shown as a green sphere, and water ligands are shown as redmetal ligands are indicated with solid lines. (b) Details of FMNbinding site.

central twisted six-stranded β-sheet sandwiched byα-helices. The topology of this domain is a modifiedRossmann fold where the fifth β-strand of the cen-tral β-sheet is anti-parallel with the rest of thestrands (Fig. 3a). Comparison of CgFMNAT withknown protein structures using the Dali server29

indicates that the core of CgFMNAT is most similarto the members of the PAPS reductase-like proteinfamily. These include bacterial adenosine 5′-phos-phosulfate (APS) reductase30 [Protein Data Bank(PDB) ID 2goy; Z-score of 15.4 and rmsd of 2.5 Å for174 superimposed Cα atoms], PAPS reductase31

(1sur; Z-score of 14.0 and rmsd of 2.4 Å for 159superimposed Cα atoms), and ATP sulfurylase32

(1zun; Z-score of 13.1 and rmsd of 2.9 Å for 160superimposed Cα atoms). Similar to CgFMNAT,these proteins also have an N-terminal domain of amodified Rossmann-fold topology (Fig. 3b). The C-terminal domain of CgFMNAT is composed largely

tails of the Mg2+-ATP binding site. Mg2+-AMPCPP in theteracting with bound substrates are shown as sticks. Mg2+

spheres. Hydrogen bonds are shown as broken lines, andbinding site. (c) Stereo view of the FAD and PPi product


of loops (64%) interlaced by α6–α9 helices and twoshort 310-helices (Fig. 3a). This domain is muchlonger than those of APS and PAPS reductases andappears to be uniquely elaborated and expanded inyeast FMNAT (Supplementary Fig. S2). As will bediscussed later, residues from this domain are alsoinvolved in interactions with ATP.

Mg2+-ATP binding site

The structures of the CgFMNAT–ATP binarycomplex and the substrate ternary complex revealdetails about the interactions between the enzymeand the ATP substrate. Five structural motifs areidentified to be involved in ATP and Mg2+ binding(Fig. 3a). The PP-loop motif, extending from β1 tothe N-terminus of α3, has the sequence of60SYNGGKDC67 and is generally conserved in thesuperfamily. The ADE motif, named for ADEninebinding, corresponds to the LDTG motif in APSreductase30 and consists of a short stretch of fourresidues (107FIDH110) following strand β2. The firstarginine-containing ARG1 motif (named Arg-loop inChartron et al.30) is located in the loop connecting β4to β5 and consists of residues 163GIRHTD168. The γ-phosphate motif encompasses part of α7 and thefollowing hairpin loop from the C-terminal domainsituated above the nucleotide binding site. Near theC-terminus of the protein, another arginine-richmotif, ARG2, of sequence 296ERAGR300 is also in-volved in nucleotide binding.The ATP nucleotide binds in a crevice formed

between β1 and β4, with its βγ-phosphate tailpositioned in an anion-binding pocket near theN-terminus of α3 (Fig. 4a). A hydrogen bond net-work is formed between ATP and residues from thefive motifs (Fig. 4a). The adenine N1 nitrogen andN6 amino groups are hydrogen bonded to the main-chain amide and carbonyl of Ile108 (part of the ADE

motif), respectively. The N3 group of adenine formsa hydrogen bond with the side chain of Ser60 of thePP-loop motif, while the O2′ hydroxyl of ATPribose interacts with the main-chain carbonyl ofSer60. The O2′ hydroxyl forms another hydrogenbond with the main-chain amide of Gly163 of theARG1 motif. The PP-loop motif residues interactextensively with the phosphates of ATP throughboth main-chain and side-chain moieties. The sidechain of Asn62 interacts with the α-phosphate; themain-chain amide of Cys67 interacts with the β-phosphate; and the side chain of Lys65 provides twohydrogen bonds to the γ-phosphate. Additionalhydrogen bonds to the γ-phosphate are formedwith the Tyr216 hydroxyl and the Leu223main-chainamide. Both residues are from the γ-phosphate motif(Fig. 4a).In the presence of magnesium ions, as in the case

of the substrate ternary complex structure, theAMPCPP phosphate tail adopts two distinct confor-mations (conformers I and II; Figs. 2b and 5) indifferent monomers of the six CgFMNAT substrateternary complexes in the asymmetric unit, both ofwhich differ from that observed in the Mg2+-freeATP binary complex structure (Fig. 5). The largestdifference between these conformations is in theposition of the β-phosphate, which moves by 5.1 Åbetween conformers I and II, while the α- and γ-phosphates move by 1.6 Å and 2.2 Å, respectively. Inboth conformers, the bound Mg2+ remains in thesame position and maintains a six-ligand octahedralconfiguration (Fig. 5). In the first conformer, Mg2+ isliganded to the β- and γ-phosphate oxygens, theAsp66 side-chain carboxyl, and the three conservedwater molecules w2, w3, and w4 (Fig. 4a). W2 andw3 are further coordinated to the carboxylate groupof Asp168 of the ARG1 motif. In the second con-former, all three AMPCPP phosphoryl moieties arecoordinated to Mg2+, with the β-phosphate oxygen

Fig. 5. Stereo view of the super-position of bound ATP, Mg2+, FMN,and FAD. The dual conformationsof AMPCPP (light green) in the sub-strate ternary complex are markedas I and II. The ATP molecule in theMg2+-free binary complex structureis shown as a blue thin line. Thesubstrate FMN is shown in pink,while the product FAD is shown byatom types, with carbon atomsshown in yellow. Mg2+ (greensphere) and the correspondingwater ligands (red spheres) arealso shown.


substituting for the w4 water ligand. The β-phosphate in this conformation also interacts withArg279 of the ARG2 motif through a bifurcatedhydrogen bond (data not shown). In the Mg2+-freeATP binary complex structure, the position of the β-phosphate largely overlaps with the Mg2+ bindingsite (Fig. 5).Compared to other PAPS reductase-like proteins,

three of the five ATP-binding motifs in theCgFMNAT, PP-loop, ADE, and ARG1 motifs areconserved (Fig. S2). Although the γ-phosphate motifis also present inmembers of the PAPS reductase-likefamily such as APS reductase and bacterial ATPsulfurylase, its role in substrate binding and catalysisin these enzymes is not clear. Unique to CgFMNATandAPS reductase is the presence of the ARG2motif(Supplementary Fig. S2). In APS reductase, Arg242and Arg245 near the protein C-terminus, corre-sponding to Arg297 and Arg300 of CgFMNAT,

Fig. 6. Adenylyl transfer mechanism for CgFMNAT. (a) Iniversus 1/[ATP] at fixed FMN concentrations. Inset: The same d[ATP]. (b) Initial rates represented by the Lineweaver–Burk ploThe same data represented by a hyperbolic plot (inset) as a functhe velocity equation for an ordered bi–bi (Eq. (1)) reactant sy

provide important interactions with the phosphosul-fate group of APS. Our preliminary kinetic analysisof the R297Amutant shows that the apparent Km,ATPand Km,FMN values of the mutant increased by ∼5and ∼3 times, respectively, compared to the wild-type enzyme (data not shown), indicating that thisArg residue is indeed involved in substrate binding,presumably through interactions with the phos-phoryl groups of the substrates.

FMN/FAD binding site

Unexpectedly, the flavin binding site, as revealedin the substrate and product ternary complexes, islocated on the same side of the central β-sheet asthe adenosine moiety of ATP, where a deep troughis formed between the face of the β-sheet, helix α5,and the loop connecting the anti-parallel β5 to β6(Figs. 3 and 4). This pocket forms a unique binding

tial rates represented by the Lineweaver–Burk plot of 1/νata represented by a hyperbolic plot (inset) as a function oft of 1/ν versus 1/[FMN] at fixed ATP concentrations. Inset:tion of FMN. The rates in (a) and (b) were globally fitted tostem. (c) Proposed catalytic mechanism of CgFMNAT.

Fig. 7. Electrostatic surface potential representation ofthe CgFMNAT substrate-binding pocket. The electrostaticpotential is color ramped from −5 kT/e (red) to +5 kT/e(blue). FMN and AMPCPP are represented by sticks, andMg2+ is shown as a green sphere.


site for the flavin isoalloxazine ring that is differentfrom those observed in any other FMN- or FAD-binding protein.33 Residues from a broad range ofstructural elements are involved in the interactionwith the isoalloxazine ring (Figs. 3a and 4c). Theseinclude Met143 and Phe147 from helix α5, Ile160and Ile162 from strand β4; Asp181, Trp184, andPhe187 from the loop connecting β5 to β6, andArg189 from strand β6. These residues are highlyconserved among eukaryotic FMNATs (Supplemen-tary Fig. S3) and are collectively referred to as theflavin motif. The isoalloxazine ring is sandwichedbetween the indole ring of Trp184 and the planarguanidinium group of Arg189. Deeply buried in theflavin-binding pocket is the hydrophobic dimethyl-benzene moiety of the isoalloxazine ring, formingvan der Waals contacts with the hydrophobic sidechains of Met143, Phe147, Ile160, Ile162, andPhe187 (Fig. 4b and c). The hydrophilic lumazineside of the ring forms two specific hydrogen bondswith the enzyme, between its C4 carbonyl and themain-chain amide of Asp181, and between its N3amide and the side chain of Asp 181, respectively.In both substrate and product ternary complexstructures, the isoalloxazine group is also in van derWaals contact with the adenosine moiety of eitherAMPCPP or FAD, suggesting that the adenosinegroup is part of the isoalloxazine-binding pocket.In contrast to extensive interactions with the iso-

alloxazine ring, there are few contacts between theenzyme and the FMN phosphoribityl tail (Fig. 4b).As a result, the conformation of this part of thesubstrate is not well defined, as indicated by high B-factors, partial occupancy, and discontinuous den-sity (Fig. 2b). In five of the six monomers in theasymmetric unit, the FMN phosphoribityl tail isseen pointing toward the solvent, positioned overthe N-terminus of helix α5 and away from thebound AMPCPP, probably due to electrostatic re-pulsion (Figs. 4b and 5). This observed conformationof FMN is apparently not in the catalytically readystate, as the phosphate is too far from the boundAMPCPP. Clearly, the flexibility of FMN phosphori-bityl tail would allow it to adopt multiple confor-mations and to move close to ATP so that theadenylyltransferase reaction could occur. Compar-ing the substrate and product ternary complexes, thepositions of the isoalloxazine ring and the adenosinemoieties remain essentially unchanged (Fig. 5). Thephosphoribityl group of the product FAD becomeswell ordered when covalently linked to the adenylylgroup of ATP. The position of the PPi product is alsowell defined in the crystal structure and is identicalwith that of the βγ-phosphates of AMPCPP in con-former I. TheMg2+ in the product ternary complex iscoordinated to the PPi, the adenylyl phosphate ofFAD, Asp66, and two water molecules in a confi-guration with features of both AMPCPP conformersin the substrate ternary complex (Fig. 5). Arg297from the C-terminal ARG2 motif is found to interactwith the diphosphate moiety of FAD (Fig. 4c), sup-porting its potential involvement in binding thephosphate groups of both ATP and FMN substrates

and in positioning them for the adenylyl transferreaction.

Steady-state kinetic properties of CgFMNAT

The steady-state kinetic parameters of CgFMNATwere determined using a continuous coupled assay.The Lineweaver–Burk reciprocal plots of 1/ν versus1/[FMN] and 1/ν versus 1/[ATP] (Fig. 6a and b) areconsistent with an ordered bi–bi system,34 with thesubstrate-binding order being ATP first, followedby FMN. The same mechanism and substrate-binding order were also proposed for rat liver andbacterial FMNATs in early studies.20,21 The initialrates of the reaction were globally fitted to thegeneral equation for an ordered steady-state bireac-tant model (insets of Fig. 6a and b). The steady-statekinetic parameters obtained for CgFMNAT are asfollows: Km for ATP=10.7 ± 2.3 μM, Km forFMN=0.76±0.15 μM, and kcat=0.087 s− 1. The Km,

FMN value for CgFMNAT is similar to that obtainedfor the human FMNAT isoform II in a recent study(apparent Km,FMN of 0.36±0.06 μM), while kcatappears to be more than 10 times higher than thatof the human enzyme (0.0036±0.0001 s− 1).35 Thesevalues are somewhat different from those obtainedfor the endogenous rat liver FMNAT, where theapparent Km values for ATP and FMN are 71 μMand 9.1 μM, respectively, and Vmax is 345 nmolFAD/min/mg protein, corresponding to a kcat of0.15 s− 1.21

Proposed mechanism for CgFMNAT

The four high-resolution structures of CgFMNAT(apo form, complexe with ATP, complexe with thesubstrate FMN+AMPCPP, and complexe with theproduct FAD+PPi), along with kinetic data, allow


us to envision events in the enzyme active siteduring catalysis (Fig. 6c; Supplementary Movie). Inthis process, ATP preferably binds first to theenzyme because its binding pocket would bepartially blocked by FMN, which binds at a sitecloser to the surface (Fig. 7). Additionally, bindingATP first may help to properly position the FMNsubstrate, as the C8M methyl group of the isoallox-azine ring is packed against the adenosine moiety ofthe bound ATP. These observations are consistentwith the substrate-binding order deduced from thekinetic data. The binding of ATP induces a smalladjustment (0.3–0.5 Å) of several surroundingresidues, including Asn62, Lys65, Asp66, Ile108,and Asp168, presumably to optimize their interac-tions with Mg2+ and ATP. In the presence of Mg2+,the phosphate tail of ATP can adopt either of the twodiscrete conformations I or II, in which the Mg2+

position remains the same (Fig. 5). Upon subsequentbinding of FMN, side chains of several residuesaround the isoalloxazine ring (e.g., Met143, Phe147,Asp181, and Trp184) also make small adjustments tooptimally interact with the substrate. Due to lack ofinteraction between the enzyme and the phosphor-ibityl tail of FMN, this part of the FMN substrate ishighly flexible and able to adopt multiple conforma-tions. For the adenylation reaction to occur, the FMNphosphate would move close to the α-phosphate ofATP for the ensuing nucleophilic attack. Thepresence of Mg2+ and interaction with Arg297 (andpotentially Arg300) of the ARG2 motif may help toovercome electrostatic repulsion between the phos-phate groups of the two substrates and to positionFMN phosphate for the attack on the α-phosphate ofATP (Fig. 6c). The cleavage of the αβ-phosphodie-ster bond is facilitated by coordination of Mg2+,which is required for the reaction. The Mg2+-ATP inthe conformer I position appears to be the catalyti-cally competent conformation, which allows theFMN phosphate group to approach the α-phosphatefrom the direction opposite the β-phosphate for thedirect in-line nucleophilic attack. Minimal structuralrearrangements are observed after product forma-tion. The leaving diphosphate group is practically inthe same position as the βγ-phosphates of confor-mer I of the nucleotide and interacts with the sameset of protein residues and Mg2+. The transferred α-phosphate moves by about 2.5 Å away from itsoriginal position and is now directly liganded toMg2+ (Fig. 5; Supplementary Movie). Earlier kineticstudies of rat liver FMNAT revealed that the enzymeis markedly inhibited by the product FAD with a Kiof 0.75 μM against FMN and a Ki of 1.3 μM againstMg2+-ATP.21 It has been suggested that the biosynth-esis of FAD is most likely regulated by the productFAD at the last FMNAT step of the pathway.21

Although not yet determined explicitly, our obser-vation that CgFMNAT copurifies with intrinsicallybound FAD, along with extensive interactionsbetween the enzyme and FAD seen in the crystalstructure, suggests that FAD binds to CgFMNATwith high affinity and likely exerts a feedbackinhibitory effect as well.

Discussion

Our high-resolution structures of yeast FMNATpresent the first characterization of a eukaryoticversion of this essential enzyme. Although the over-all structure of CgFMNATshares a significant simila-rity with the PAPS-reductase-like family of proteins,as reflected in several shared nucleotide-bindingmotifs, the mode of flavin binding in CgFMNAT hasnot been observed before. In a 2001 survey by Dymand Eisenberg, all FAD-binding proteins withknown three-dimensional structures were catego-rized into four different groups, represented byglutathione reductase, ferredoxin reductase, p-cresolmethylhydroxylase, and pyruvate oxidase.33

CgFMNAT is clearly distinct from any of theseFAD-binding proteins in both the conservedsequence motifs and the flavin-binding mode. Inother Rossmann-fold FAD-binding proteins (e.g.,glutathione reductase and pyruvate oxidase groups),the isoalloxazine ring of FAD invariably binds onthe side of the central β-sheet opposite from theadenine moiety and across the top of the sheetbetween β1 and β4. It often interacts with residuesfrom another domain. In contrast, in CgFMNAT,the isoalloxazine ring binds to the same side of thecentral β-sheet as the adenine, and the bindingexclusively involves residues from the Rossmann-like N-terminal domain. The deviation from thetypical Rossmann-fold topology, characterized byan anti-parallel strand β5 at the edge of the β-sheet,opens up the side of the β-sheet to form the flavin-binding pocket. In the CgFMNAT product ternarycomplex structure, FAD adopts a bent conforma-tion with adenosine moiety and isoalloxazine ringpack against each other. While the FAD cofactor inmost flavoproteins adopts an extended conforma-tion, bent FAD has been observed in flavodoxinreductase and DNA photolyase.33 Yet, the con-formations of these bent FADs are very differentfrom that in CgFMNAT (Fig. 8), further empha-sizing the unique flavin-binding mode of eukar-yotic FMNAT.A comparison of the structures of CgFMNAT, a

prototypical eukaryotic FMNAT, and bacterialFMNAT, as exemplified by Thermotoga maritimaFADS (TmFADS),39 revealed remarkable differencesin substrate-binding modes and catalytic site con-figurations. TmFADS belongs to the large nucleoti-dylyl transferase superfamily, with the signature(H/T)xGHmotif located between the end of the firstβ-strand and the first helix of the Rossmann-foldcore. It has a conformation different from that of thecorresponding PP-loop region of CgFMNAT (Fig. 9).The second conserved motif of the nucleotidylyltransferase superfamily ISSTxxR is located at the N-terminal end of an α-helix in a C-terminal sub-domain and interacts with the β- and γ-phosphatesof the ATP nucleotide. No equivalent structuralmotif corresponding to the ISSTxxR motif exists inCgFMNAT, although the γ-phosphate motif appearsto perform a similar role. Most strikingly, the bound

Fig. 8. Comparison of protein-bound FAD conformations. Representative protein-bound FAD molecules weresuperimposed over the isoalloxazine rings. FAD bound to CgFMNAT (yellow) is compared to FAD from (a) flavodoxinreductase (blue) (PDB ID 1fdr);36 (b) DNA photolyase (magenta) (PDB ID 1dnp);37 and (c) glutathione reductase (green)(PDB ID 3grs).38


adenine nucleotides in the two proteins are orientedin opposite directions with regard to the plane of thecentral β-sheet (Fig. 9), delineating two completelydifferent nucleotide-binding modes in these twoprotein superfamilies. Currently, no flavin-boundbacterial FMNATstructure is available. Based on thestructural similarity of TmFADS to other members ofnucleotidylyl transferases such as nicotinamidemononucleotide adenylyltransferases (NMNATs)for which extensive structural information isavailable,40–42 the FMN substrate likely binds to asite corresponding to the nicotinamide mononucleo-tide binding site in NMNAT on the opposite side ofthe central β-sheet from ATP so that the productFAD adopts a largely extended conformation (Fig. 9,right). Again, this arrangement is very differentfrom the flavin-binding mode observed inCgFMNAT (Fig. 9, left). Thus, eukaryotic and

bacterial FMNATs present a remarkable case ofancient Rossmann-fold proteins that, after firstdiverging into two distinct protein families withdifferent nucleotide-bindingmodes, have developedthe same enzymatic activity via different active-siteconfigurations.In summary, we present here high-resolution

structures of a eukaryotic FMN adenylyltransferasein different complexed states, which revealed detailsabout active-site configuration and a unique FAD-binding mode. The extensive interactions observedbetween the enzyme and the product FAD, and thelack of conformational changes after catalysis mayexplain why the enzyme is likely inhibited by theproduct. These structures lay the foundation forfuture investigation of the functional roles of active-site residues and the mechanism by which FMNATinfluences FAD homeostasis in cells.

Fig. 9. Comparison of eukaryo-tic and bacterial FMNAT structures.Ribbon diagrams of the Rossmann-like fold core of the CgFMNAT–FAD complex (left) and the FMNATdomain of TmFADS (right) areshown in roughly the same orienta-tion. A FAD molecule is modeled inthe TmFMNAT active site based onthe TmFADS–AMP complex struc-ture (PDB ID 1t6y)39 and homo-logous NMNAT–NAD complexstructures40. Corresponding struc-tural elements are colored identi-cally in the two structures.


Materials and Methods

Protein expression and purification

The predicted gene encoding CgFMNAT (gi|50291750)was amplified from C. glabrata genomic DNA (strainNCYC; ATCC36909D) by PCR and cloned into the NcoIand SalI restriction sites of the pHIS parallel vector.43 Theplasmid was transformed into Escherichia coli BL21(DE3)(Novagen), and the His6-CgFMNAT protein wasexpressed at 20 °C. His6-CgFMNAT was first loaded on anickel–Sepharose affinity column (GE Healthcare) equili-brated in buffer A [20 mM Hepes (pH 8.0), 300 mM NaCl,20 mM imidazole, 5% (vol/vol) glycerol, and 1 mM DTT]and eluted with a gradient of 20–500 mM imidazole inbuffer A. The His6-tag was cleaved by tobacco etch virusprotease during overnight dialysis at 4 °C and wasremoved from CgFMNAT by passing through the nickel–Sepharose affinity column for a second time. As a secondpurification step, protein was loaded onto a Resource Qanion exchange column (GE Healthcare) equilibrated withbuffer B [20mMHepes (pH 7.5) and 5% (vol/vol) glycerol]and elutedwith a 0−350mMNaCl gradient, which yieldedtwo pools. The first pool was in bright yellow and waslater shown to contain the CgFMNAT–FAD complex. Thesecond pool was in light yellow, indicating the presence offlavin with partial occupancy. To remove the flavin and toobtain homogeneous apo-CgFMNAT, the second pool wasincubated with 1.5 M (NH4)2SO4 and purified by phenyl–Sepharose hydrophobic interaction chromatography (GEHealthcare). The final purification step for bothCgFMNAT–FAD and apo-CgFMNAT consisted of a size-exclusion chromatography column (Superdex 75 16/60;GE Healthcare), and the protein was eluted in 20 mMHepes (pH 7.5), 150 mM NaCl, and 1 mM DTT. The sele-nomethionyl apo-CgFMNATwas expressed in E. coli BL21(DE3) grown in minimal media supplemented withselenomethionine and other nutrients, and purified usingthe same procedure as for apo-CgFMNAT.

Crystallization

All crystals were grown using hanging-drop vapor-diffusion methods. For apo-CgFMNAT and ATP complexcrystallization, the reagents and greased 24-well plateswere chilled on ice before setting up crystallization drops.The apo-CgFMNAT crystals were grown by mixing 1.5 μlof protein (24 mg/ml in the gel-filtration column buffer)with 1.5 μl of reservoir solution composed of 0.1 M Naacetate (pH 4.4–5.4) and 6–12% (wt/vol) polyethyleneglycol 4000 and by equilibrating against the reservoir at16 °C. Prism-shaped apo-CgFMNAT crystals grew to amaximum size of 0.55 mm×0.35 mm×0.35 mm withinseveral days. Selenomethionyl apo-CgFMNAT crystalswere grown under similar conditions as the native protein.The CgFMNAT–ATP complex crystals were grown in thepresence of 10 mM ATP under similar conditions as thosefor apo-CgFMNAT. The cocrystals of the CgFMNAT sub-strate ternary complex were grown at 20 °C in the pre-sence of 0.2 mM FMN and 5 mM AMPCPP under similarconditions as those for apo-CgFMNAT, except that con-secutive microseeding procedure was performed in orderto obtain single crystals. The CgFMNAT product ternarycomplex was obtained by adding 2 mM Na pyropho-sphate to the CgFMNAT–FAD pool (final protein concen-tration of∼20mg/ml in gel-filtration buffer). The complexcrystals were grown by mixing 1.5 μl of the CgFMNAT–

FAD–PPi complex with 1.5 μl of reservoir solution com-posed of 0.1 M Na acetate (pH 4.4–5.4), 0.2 MMgSO4, and20–26% (wt/vol) polyethylene glycol MME 2000, and byequilibrating against the reservoir at 20 °C. All crystalswere cryoprotected in a solution containing all the reser-voir components and increments of glucose (10%, 20%,and 30%), flash-frozen in liquid propane, and stored inliquid nitrogen. For cryoprotection of the substrate ternarycomplex crystals, 5 mMMgSO4was also added alongwithglucose.

Data collection, X-ray structure determination,and refinement

The 2.18-Å SAD data from a selenomethionyl apo-CgFMNAT crystal was collected at the absorption edge(K-edge) of selenium at beamline 19-BM; the 1.20-Å nativeapo-CgFMNATand the 1.35-Å CgFMNAT product ternarycomplex data sets were collected at beamline 19-ID of theAdvance Photon Source, Argonne National Laboratory.The 1.87-Å data set for the ATP complex and the 1.95-Ådata set for the substrate ternary complex were collectedin-house with X-ray from a rotating anode generator(Rigaku FRE SuperBright) recorded on an R-AXIS IV++

(Rigaku) image plate detector. All data were processedwith either the HKL2000 package or the HKL3000package.44,45 Data collection statistics are presented inTable 1. The initial phases for apo-CgFMNAT wereobtained by the SAD phasing method (see SupplementaryMethods for details), while the phases of all complexeswere determined by the molecular replacement methodusing the program MOLREP46 and the refined apo-CgFMNAT as search model. The refinements were per-formed using REFMAC47 or PHENIX,48 and manualmodel building was performed with Coot.49 Models wereassessed by MolProbity.50

Steady-state kinetics

Steady-state kinetic parameters were determined usinga continuous coupled assay (EnzChek® pyrophosphateassay kit; Molecular Probes), with modifications. Thereactionmixture (0.5 ml) contained 20mMHepes (pH 7.5),2 mM MgCl2, 0.5 U of purine nucleoside phosphorylase,0.015 U of inorganic pyrophosphatase, 0.2 mM 2-amino-6-mercapto-7-methylpurine riboside, 21 nM (10.5 pmol or378 ng) apo-CgFMNAT, 2.5–20 μM FMN, and 1–250 μMATP. All enzyme reactions were carried out in duplicateand performed at 25 °C. The reaction was initiated by theaddition of ATP, and progress was monitored at 360 nmfor 5 min. Steady-state kinetic parameters were deter-mined by fitting the initial rates to the general equationdescribing a bireactant ordered bi–bi system (Eq. (1)):34

V =Vmax ATP�½FMN½ �

Ki;ATPKm;FMN +Km;FMN½ATP� +Km;ATP½FMN� þ ½ATP�½FMN�ð1Þ

where Km,ATP and Km,FMN are the Michaelis–Mentenconstants, and Ki,ATP is the dissociation constant for anATP–enzyme complex. A nonlinear least-squares method,as implemented in the Sigma Plot Enzyme Kineticsmodule (Systat Software), was used to fit the data.

Accession codes

The coordinates and structures factors have been depo-sited in the PDB with accession codes 3FWK (apo-


CgFMNAT), 3G59 (CgFMNAT–ATP complex), 3G5A(CgFMNAT–FMN–AMPCPP–Mg2+ complex), and 3G6K(CgFMNAT–FAD–PPi–Mg2+ complex).

Acknowledgements

We thank Zbyszek Otwinowski for help withsynchrotron data collection and initial SAD pha-sing. This work was supported by a grant from theWelch Foundation (I-5105). C.H. was supported byNational Institutes of Health training grant T32GM008297. Results shown in this report are derivedfrom work performed at Argonne National Labo-ratory, Structural Biology Center at the AdvancedPhoton Source. Argonne National Laboratory isoperated by UChicago Argonne, LLC, for the USDepartment of Energy, Office of Biological andEnvironmental Research, under contract no. DE-AC02-06CH11357.

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2009.04.022

References

1. Massey, V. (2000). The chemical and biologicalversatility of riboflavin. Biochem. Soc. Trans. 28,283–296.

2. Müller, F. (1991). In Chemistry and Biochemistry ofFlavoenzymes (Müller, F., ed.), vol. 1, pp. 1–77, CRCPressBoca Raton, FL.

3. Joosten, V. & van Berkel, W. J. (2007). Flavoenzymes.Curr. Opin. Chem. Biol. 11, 195–202.

4. van Berkel,W. J. H. (2008). Chemistry of flavoenzymes.Wiley Encyclopedia of Chemical Biology, August 15, 2008edit. John Wiley & Sons, Inc., Hoboken, N.J.

5. Merrill, A. H., Jr, Lambeth, J. D., Edmondson, D. E. &McCormick, D. B. (1981). Formation and mode ofaction of flavoproteins. Annu. Rev. Nutr. 1, 281–317.

6. Fischer, M. & Bacher, A. (2005). Biosynthesis offlavocoenzymes. Nat. Prod. Rep. 22, 324–350.

7. Powers, H. J. (2003). Riboflavin (vitamin B-2) andhealth. Am. J. Clin. Nutr. 77, 1352–1360.

8. Vogl, C., Grill, S., Schilling, O., Stulke, J., Mack, M.& Stolz, J. (2007). Characterization of riboflavin(vitamin B2) transport proteins from Bacillus subtilisand Corynebacterium glutamicum. J. Bacteriol. 189,7367–7375.

9. Reihl, P. & Stolz, J. (2005). The monocarboxylatetransporter homolog Mch5p catalyzes riboflavin(vitamin B2) uptake in Saccharomyces cerevisiae.J. Biol. Chem. 280, 39809–39817.

10. Manstein, D. J. & Pai, E. F. (1986). Purification andcharacterization of FAD synthetase from Brevibac-terium ammoniagenes. J. Biol. Chem. 261, 16169–16173.

11. Coquard, D., Huecas, M., Ott, M., van Dijl, J. M., vanLoon, A. P. & Hohmann, H. P. (1997). Molecularcloning and characterisation of the ribC gene from

Bacillus subtilis: a point mutation in ribC results inriboflavin overproduction.Mol. Gen. Genet. 254, 81–84.

12. Santos, M. A., Jimenez, A. & Revuelta, J. L.(2000). Molecular characterization of FMN1, thestructural gene for the monofunctional flavokinaseof Saccharomyces cerevisiae. J. Biol. Chem. 275,28618–28624.

13. Wu, M., Repetto, B., Glerum, D. M. & Tzagoloff, A.(1995). Cloning and characterization of FAD1, thestructural gene for flavin adenine dinucleotidesynthetase of Saccharomyces cerevisiae. Mol. Cell. Biol.15, 264–271.

14. Liu, M. T., Wuebbens, M. M., Rajagopalan, K. V. &Schindelin, H. (2000). Crystal structure of thegephyrin-related molybdenum cofactor biosynthesisprotein MogA from Escherichia coli. J. Biol. Chem.275, 1814–1822.

15. Schrag, J. D., Huang, W., Sivaraman, J., Smith, C.,Plamondon, J., Larocque, R. et al. (2001). The crystalstructure of Escherichia coli MoeA, a protein from themolybdopterin synthesis pathway. J. Mol. Biol. 310,419–431.

16. Karthikeyan, S., Zhou, Q., Mseeh, F., Grishin, N. V.,Osterman, A. L. & Zhang, H. (2003). Crystal structureof human riboflavin kinase reveals a beta barrel foldand a novel active site arch. Structure, 11, 265–273.

17. Cheek, S., Zhang, H. & Grishin, N. V. (2002).Sequence and structure classification of kinases.J. Mol. Biol. 320, 855–881.

18. Murzin, A. G., Brenner, S. E., Hubbard, T. &Chothia, C. (1995). SCOP: a structural classificationof proteins database for the investigation ofsequences and structures. J. Mol. Biol. 247, 536–540.

19. Finn, R. D., Tate, J., Mistry, J., Coggill, P. C.,Sammut, S. J., Hotz, H. R. et al. (2008). The Pfamprotein families database. Nucleic Acids Res. 36,D281–D288.

20. Efimov, I., Kuusk, V., Zhang, X. & McIntire, W. S.(1998). Proposed steady-state kinetic mechanismfor Corynebacterium ammoniagenes FAD synthetaseproduced by Escherichia coli. Biochemistry, 37,9716–9723.

21. Yamada, Y., Merrill, A. H., Jr & McCormick, D. B.(1990). Probable reaction mechanisms of flavokinaseand FAD synthetase from rat liver. Arch. Biochem.Biophys. 278, 125–130.

22. Tu, B. P. & Weissman, J. S. (2002). The FAD- and O(2)-dependent reaction cycle of Ero1-mediated oxida-tive protein folding in the endoplasmic reticulum.Mol. Cell, 10, 983–994.

23. Hustad, S., Ueland, P. M., Vollset, S. E., Zhang, Y.,Bjorke-Monsen, A. L. & Schneede, J. (2000).Riboflavin as a determinant of plasma total homo-cysteine: effect modification by the methylenete-trahydrofolate reductase C677T polymorphism.Clin. Chem. 46, 1065–1071.

24. Gerdes, S. Y., Scholle, M. D., D'Souza, M., Bernal, A.,Baev, M. V., Farrell, M. et al. (2002). From geneticfootprinting to antimicrobial drug targets: examplesin cofactor biosynthetic pathways. J. Bacteriol. 184,4555–4572.

25. Ostrosky-Zeichner, L., Rex, J. H., Pappas, P. G.,Hamill, R. J., Larsen, R. A., Horowitz, H. W. et al.(2003). Antifungal susceptibility survey of 2,000bloodstream Candida isolates in the United States.Antimicrob. Agents Chemother. 47, 3149–3154.

26. Krcmery, V. & Barnes, A. J. (2002). Non-albicansCandida spp. causing fungaemia: pathogenicity andantifungal resistance. J. Hosp. Infect. 50, 243–260.


27. Merritt, E. A. (1999). Expanding the model: aniso-tropic displacement parameters in protein structurerefinement. Acta Crystallogr. Sect. D, 55, 1109–1117.

28. Potterton, E., Briggs, P., Turkenburg, M. &Dodson, E. (2003). A graphical user interface tothe CCP4 program suite. Acta Crystallogr. Sect. D,59, 1131–1137.

29. Holm, L. & Sander, C. (1995). Dali: a network toolfor protein structure comparison. Trends Biochem.Sci. 20, 478–480.

30. Chartron, J., Carroll, K. S., Shiau, C., Gao, H., Leary,J. A., Bertozzi, C. R. & Stout, C. D. (2006). Substraterecognition, protein dynamics, and iron–sulfur clus-ter in Pseudomonas aeruginosa adenosine 5′-phospho-sulfate reductase. J. Mol. Biol. 364, 152–169.

31. Savage, H., Montoya, G., Svensson, C., Schwenn,J. D. & Sinning, I. (1997). Crystal structure ofphosphoadenylyl sulphate (PAPS) reductase: a newfamily of adenine nucleotide alpha hydrolases.Structure, 5, 895–906.

32. Mougous, J. D., Lee, D. H., Hubbard, S. C., Schelle,M. W., Vocadlo, D. J., Berger, J. M. & Bertozzi, C. R.(2006). Molecular basis for G protein control of theprokaryotic ATP sulfurylase. Mol. Cell, 21, 109–122.

33. Dym, O. & Eisenberg, D. (2001). Sequence–structureanalysis of FAD-containing proteins. Protein Sci. 10,1712–1728.

34. Segel, I. H. (1993). Enzyme Kinetics: Behavior andAnalysis of Rapid Equilibrium and Steady-State EnzymeSystems. Wiley-Interscience, New York.

35. Galluccio, M., Brizio, C., Torchetti, E. M., Ferranti,P., Gianazza, E., Indiveri, C. & Barile, M. (2007).Over-expression in Escherichia coli, purification andcharacterization of isoform 2 of human FADsynthetase. Protein Expression Purif. 52, 175–181.

36. Ingelman, M., Bianchi, V. & Eklund, H. (1997). Thethree-dimensional structure of flavodoxin reductasefrom Escherichia coli at 1.7 Å resolution. J. Mol. Biol.268, 147–157.

37. Park, H. W., Kim, S. T., Sancar, A. & Deisenhofer, J.(1995). Crystal structure of DNA photolyase fromEscherichia coli. Science, 268, 1866–1872.

38. Karplus, P. A. & Schulz, G. E. (1987). Refined structureof glutathione reductase at 1.54 Å resolution. J. Mol.Biol. 195, 701–729.

39. Wang, W., Kim, R., Yokota, H. & Kim, S. H. (2005).Crystal structure of flavin binding to FAD synthetaseof Thermotoga maritima. Proteins, 58, 246–248.

40. Saridakis, V., Christendat, D., Kimber, M. S.,Dharamsi, A., Edwards, A. M. & Pai, E. F. (2001).Insights into ligand binding and catalysis of a centralstep in NAD+ synthesis: structures of Methanobac-terium thermoautotrophicum NMN adenylyltransferasecomplexes. J. Biol. Chem. 276, 7225–7232.

41. Zhang, H., Zhou, T., Kurnasov, O., Cheek, S.,Grishin, N. V. & Osterman, A. (2002). Crystal struc-tures of E. coli nicotinate mononucleotide adenylyl-transferase and its complex with deamido-NAD.Structure (Cambridge), 10, 69–79.

42. Zhou, T., Kurnasov, O., Tomchick, D. R., Binns, D. D.,Grishin, N. V., Marquez, V. E. et al. (2002). Structure ofhuman nicotinamide/nicotinic acid mononucleotideadenylyltransferase. Basis for the dual substratespecificity and activation of the oncolytic agenttiazofurin. J. Biol. Chem. 277, 13148–13154.

43. Sheffield, P., Garrard, S. & Derewenda, Z. (1999).Overcoming expression and purification problemsof RhoGDI using a family of “parallel” expressionvectors. Protein Expression Purif. 15, 34–39.

44. Otwinowski, Z. & Minor, W. (1997). Processing ofX-ray diffraction data collected in oscillation mode.Methods Enzymol. 276, 307–326.

45. Minor, W., Cymborowski, M., Otwinowski, Z. &Chruszcz, M. (2006). HKL-3000: the integration ofdata reduction and structure solution—from diffrac-tion images to an initial model in minutes. ActaCrystallogr. Sect. D, 62, 859–866.

46. Vagin, A. & Teplyakov, A. (1997). MOLREP: an auto-mated program for molecular replacement. J. Appl.Crystallogr. 30, 4.

47. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997).Refinement of macromolecular structures by themaximum-likelihood method. Acta Crystallogr. Sect.D, 53, 240–255.

48. Adams, P. D., Grosse-Kunstleve, R. W., Hung, L. W.,Ioerger, T. R., McCoy, A. J., Moriarty, N. W. et al.(2002). PHENIX: building new software for auto-mated crystallographic structure determination. ActaCrystallogr. Sect. D, 58, 1948–1954.

49. Emsley, P. & Cowtan, K. (2004). Coot: model-buildingtools for molecular graphics. Acta Crystallogr. Sect. D,60, 2126–2132.

50. Lovell, S. C., Davis, I. W., Arendall, W. B., III, deBakker, P. I., Word, J. M., Prisant, M. G. et al. (2003).Structure validation by Calpha geometry: phi, psiand Cbeta deviation. Proteins, 50, 437–450.

Documents

Structure and Mechanism of a Eukaryotic FMN Adenylyltransferase