the crystal structure of a ternary complex of betaine ... · molecular y bioprocesos, instituto de...

doi:10.1016/j.jmb.2008.10.082 J. Mol. Biol. (2009) 385, 542–557

Available online at www.sciencedirect.com

The Crystal Structure of A Ternary Complex of BetaineAldehyde Dehydrogenase from Pseudomonasaeruginosa Provides New Insight into the ReactionMechanism and Shows A Novel Binding Mode of the2′-Phosphate of NADP+ and A Novel Cation Binding Site

Lilian González-Segura1†, Enrique Rudiño-Piñera1†,Rosario A. Muñoz-Clares2* and Eduardo Horjales1*

1Departamento de MedicinaMolecular y Bioprocesos,Instituto de Biotecnología,Universidad NacionalAutónoma de México, Av.Universidad 2001, Cuernavaca,Morelos CP 62250, Mexico2Departamento de Bioquímica,Facultad de Química,Universidad NacionalAutónoma de México, MéxicoDF 04510, Mexico

Received 20 August 2008;received in revised form10 October 2008;accepted 22 October 2008Available online5 November 2008

*Corresponding authors. E-mail addPresent address: L. González-Segu

Autónoma de México, México DF 04† L. González-Segura and E. RudAbbreviations used: ALDH, aldeh

nonphosphorylating glyceraldehydebetaine ALDH from Pseudomonas aethermophilus.

0022-2836/$ - see front matter © 2008 E

In the human pathogen Pseudomonas aeruginosa, the NAD(P)+-dependentbetaine aldehyde dehydrogenase (PaBADH) may play the dual role ofassimilating carbon and nitrogen from choline or choline precursors—abundant at infection sites—and producing glycine betaine and NADPH,potentially protective against the high-osmolarity and oxidative stressesprevalent in the infected tissues. Disruption of the PaBADH gene negativelyaffects the growth of bacteria, suggesting that this enzyme could be a target forantibiotic design. PaBADH is one of the few ALDHs that efficiently useNADP+ and one of the even fewer that require K+ ions for stability. Crystals ofPaBADH were obtained under aerobic conditions in the presence of 2-mercaptoethanol, glycerol, NADP+ and K+ ions. The three-dimensionalstructure was determined at 2.1-Å resolution. The catalytic cysteine (C286,corresponding to C302 of ALDH2) is oxidized to sulfenic acid or forms amixed disulfide with 2-mercaptoethanol. The glutamyl residue involved inthe deacylation step (E252, corresponding to E268 of ALDH2) is in twoconformations, suggesting a proton relay system formed by two well-conserved residues (E464 and K162, corresponding to E476 and K178,respectively, of ALDH2) that connects E252 with the bulk water. In someactive sites, a bound glycerol molecule mimics the thiohemiacetal inter-mediate; its hydroxyl oxygen is hydrogen bonded to the nitrogen of the amidegroups of the side chain of the conservedN153 (N169 of ALDH2) and those ofthe main chain of C286, which form the “oxyanion hole.” The nicotinamidemoiety of the nucleotide is not observed in the crystal, and the adeninemoietybinds in the usual way. A salt bridge between E179 (E195 of ALDH2) and R40(E53 of ALDH2)moves the carboxylate group of the former away from the 2′-phosphate of the NADP+, thus avoiding steric clashes and/or electrostaticrepulsion between the two groups. Finally, the crystal shows two K+ bindingsites per subunit. One is in an intrasubunit cavity that we found to be presentin all known ALDH structures. The other\not described before for anyALDH but most likely present in most of them\is located in between thedimeric unit, helping structure a region involved in coenzyme binding and

resses: [email protected]; [email protected], Departamento de Bioquímica, Facultad de Química, Universidad Nacional510, Mexico.iño-Piñera contributed equally to this work.yde dehydrogenase; FDH, formyltetrahydrofolate dehydrogenase; GAPN,-3-phosphate dehydrogenase; NMN, nicotinamide mononucleotide; PaBADH,ruginosa; TtP5CDh, Δ1-pyrroline-5-carboxylate dehydrogenase of Thermus

lsevier Ltd. All rights reserved.

543Three-Dimensional Structure of Betaine Aldehyde Dehydrogenase

catalysis. Thismay explain the effects of K+ ions on the activity and stability ofPaBADH.

© 2008 Elsevier Ltd. All rights reserved.

Keywords: betaine aldehyde dehydrogenase three-dimensional structure;cysteine oxidation; proton relay system; ternary complex; glycerol, NADP+

and K+ ion binding
Edited by R. Huber
‡The amino acid numbering of ALDH2 is given withinparentheses next to the PaBADH numbering to facilitate acomparison between the structure of PaBADH and thestructures of other ALDHs.

Introduction

In a wide variety of prokaryotic and eukaryoticorganisms, the enzyme betaine aldehyde dehydro-genase (ALDH) [betaine aldehyde: NAD(P)+ oxidor-eductase, E.C. 1.2.1.8, BADH] catalyzes theirreversible NAD(P)+-dependent oxidation of betainealdehyde, producing the osmoprotectant glycinebetaine.1–3 The reaction catalyzed by BADH consti-tutes the second step of the choline catabolic pathwayin certain bacteria that are able to grow in choline orcholine precursors as their only carbon, nitrogen andenergy sources, such as the human pathogen Pseudo-monas aeruginosa.4,5 Inhibition of the enzyme from P.aeruginosa (PaBADH) would not only block or reducecholine catabolism or the supply of glycine betainebut also lead to the accumulation of betaine aldehyde,a highly toxic aldehyde.6 This enzyme might there-fore be a target for amuchneeded antibiotic, given thehigh prevalence of antibiotic-resistant strains of P.aeruginosa and the susceptibility to infection of agrowing immunodepressed population. Validation ofPaBADH as a suitable target for antimicrobial agentscomes from the finding that disruption of thePaBADH gene severely affects the growth of thebacterium on glucose if osmotic stress conditions andcholine are also present in the medium.6 Given thatthe nucleophilic attack of an essential cysteine residueon the aldehyde substrate is the first step in thecatalytic mechanism of every ALDH, thiol-specificreagents have been tested as potential inhibitors ofPaBADH activity and P. aeruginosa growth.7,8 Forrational development and/or optimization of clini-cally useful inhibitors, however, we require knowl-edge of the enzyme structure and its reactionmechanism.The chemical mechanism of the reaction catalyzed

by ALDHs involves two nucleophilic attacks, oneoccurring in the acylation step (i.e., formation of thethiohemiacetal intermediate, carried out by the thiolgroup of the catalytic cysteine) and the other in thedeacylation step (i.e., hydrolysis of the thioesterintermediate, carried out by a water molecule). Bothnucleophiles need to be “activated” by lowering theirhigh pKa so they can lose a proton producing thereactive thiolate and hydroxyl ion, respectively, atphysiological pH. In the vicinity of the catalyticcysteine, there are two highly conserved glutamateresidues (ALDH2 numbering, E268 and E398;PaBADH numbering, E252 and E387), each of whichhas been proposed to be the general base in thereaction catalyzed by different ALDHs.9–11 In the caseof PaBADH, it is not known which of these two

glutamyl residues participates in the activation of thereaction nucleophiles.Unlike most ALDHs that prefer NAD+, PaBADH

uses NADP+ with the same efficiency as NAD+,12

even though sequence alignments13 have shown thatit has the glutamyl residue (ALDH2numbering, E195;PaBADH numbering, E179) for long thought to beincompatible with the binding of the 2′-phosphate ofNADP+ and therefore characteristic of NAD+-depen-dent ALDHs.14 The only three-dimensional structureof a BADH so far determined, that of the NAD+-dependent cod liver enzyme,15 suggests that a prolineat a position equivalent to that of E179(E195)‡wouldproduce a steric clash with the 2′-phosphate, thusaccounting for its low affinity for NADP+. The three-dimensional structures of ALDHs that can bindNADP+ with similar or higher affinity than NAD+

have shown different strategies to accommodate the2′-phosphate group.16–20 From analysis of thePaBADH amino acid sequence, however, as well asthat of the amino acids known to be involved inbinding of NADP+, it is not clear which of thesedescribed modes, if any, applies in the case ofPaBADH.We also do not know the structural bases of the

stabilizing effects that monovalent cations, particu-larly K+ ions, have on PaBADH.21–23 In the absence ofK+ ions, PaBADH loses activity in a time-dependentmanner, which is caused by enzyme dissociation.Other BADHs—such as those from Escherichia coli,24

amaranth leaves,25 Bacillus subtilis,26 porcine kidney27

and horseshoe crab28—were found to be activated byK+ ions to some extent, but of these, only the stabilityof the kidney BADH has been reported to be affectedby monovalent cations.22 Regarding the ALDHsuperfamily, the possible role of monovalent cationson the activity and stability of these enzymes has notbeen extensively studied, and this may be the reasonthat the only examples to date ofK+-activatedALDHsare an ethanol-inducible enzyme from P. aeruginosa29

and two mitochondrial ALDH isoenzymes fromSaccharomyces cerevisiae.30,31 The structure of any ofthese K+-dependent ALDHs has so far not beendetermined, nor have the ligands of the K+ bindingsites been identified. The human ALDH2 has a Na+

ion bound per subunit,32 but its functional orstructural relevance is still unknown.

Table 1. Data collection and refinement statistics forPaBADH

Crystal PaBADH-NADP+

Crystal parametersSpace group C121Cell dimensionsa, b, c (Å) 334.9, 133.0, 101.8α, β, γ (°) 90.0, 94.9, 90.0Asymmetric unit Two tetramers

RefinementResolution (Å)a 44.9–2.10b

Unique reflections 233,530Completeness (%) 93.0 (79.3)

544 Three-Dimensional Structure of Betaine Aldehyde Dehydrogenase

Here, we present the crystal structure ofPaBADH\complexed with glycerol, NADP+ and K+

ions\determined at a resolution of 2.1 Å, revealing anovel mode of binding of the 2′-phosphate group ofthe NADP+ and a novel intersubunit K+ ion bindingsite, as well as an intrasubunit K+ ion binding sitesimilar to that found in ALDH2.32 These cationbinding sites appear to be a general feature ofALDHs. In addition, the structure shows the catalyticcysteine modified to sulfenic acid or forming a mixeddisulfide, a glycerol molecule bound to some activesites mimicking the thiohemiacetal intermediate and alikely proton relay system that participates in catalysis.

I/σ (I)–Mean I/σ (I) 8.2 (1.6)–12.5 (2.4)Rmerge

c (%) 7.7 (44.9)Rwork

d (%) 16.6Rfree

e (%) 21.1Protein atoms 30,135Ligand atoms 418Water atoms 2932RMSD from ideal stereochemistryBond lengths (Å) 0.016Bond angles (°) 1.535Mean B values (Å2)Mean B-value (Protein) 22.9f

Mean B-value (Solvent) 27.2Mean B-value (Ligand) 35.5Mean overall B-value (Å2)Overall 21.7f

Model qualityRamachandran plot (%)Most favored 96.5Additional allowed 3.5a Values in parentheses correspond to the highest-resolution

shell.b Even though the data were, in general, statistically sound at

2.0-Å resolution, the refinement was carried out at the highestresolution of 2.1 Å, stressing the importance of Rmerge and I/σ(I)values at that resolution shell [between 2.1 and 2.2 Å;Rmerge=30.1% and I/σ(I)=2.4].

c Rmerge=∑j∑h(|Ij,h− ⟨Ih⟩|)/∑j∑h(⟨Ih⟩), where h is the uniquereflection index, Ij,h is the intensity of the symmetry-relatedreflection and ⟨Ih⟩ is the mean intensity.

d Rwork=∑h||Fo|h−|Fc|h|/∑h|Fo|h, where h defines theunique reflections.

e Calculated on a random, but constant, 5% of the data.f Residual B values are given; the others were considered into

the calculated TLS tensor and used in the refinement process.

Results and Discussion

Structure determination and maincharacteristics

The structure of PaBADH complexed with NADP+

has been determined at 2.1-Å resolution. The crystalbelongs to the C121 space group and contains eightsubunits in the asymmetric unit describing twotetramers, each ofwhich corresponds to the biologicalunit of the enzyme.22 The solution of the molecularreplacement showed that seven of the eight subunitsclearly localized. The eighth subunit (labeled assubunit H)\whose electron density was remarkablyweaker compared with the other subunits\wasinitially built by using the noncrystallographic sym-metries of the well-defined tetramer. After buildingthe first trace, however, long fragments of the aminoacid sequence were erased because of lack of electrondensity, and only after several cycles of building andrebuilding was the full sequence fitted. When each ofthe α carbons of the eight PaBADH subunits wascompared with its corresponding α carbon in subunitA of cod BADH, the RMSD values were within therange of 0.87–0.95 Å. Some regions presented higherRMSD values, but they were always less than 2.5 Å.The final model has an R value of 16.6% and an Rfreevalue of 21.1%. All data collection and refinementstatistics are summarized in Table 1.PaBADH crystals, as are the rest of the tetrameric

ALDH examples deposited in the Protein Data Bank(PDB), are builtwith tetramers that can bedescribedasa dimer of dimers. By refining the TLS parameters,33

we could define three regions that can be consideredas quasi-rigid bodies and that superpose with thedefined and accepted catalytic nucleotide binding andoligomerization domains of ALDHs (Fig. 1). TLSanalysis also shows the different mobilities of thesubunits forming a dimer; a “less flexible” subunitalways oligomerizes with a “more flexible” subunit,with the extreme cases being subunits G and H. TheTLS refinement of the crystallographic data improvedthe statistics of the refinement, stressing the existenceof a dynamic component in the PaBADH structure.The understanding of the dynamics in PaBADHcouldhave important implications for the understanding ofits stability and catalytic mechanism.

TheNADP+molecules, one for each of the subunits,were built into the structure, but we were not able todetect the nicotinamide mononucleotide (NMN+)moiety in the electron density map. The ADP moietywas seen with occupancy factors between 0.7 and 1.0,except in subunitH,which has an occupancy factor of0.5. This low occupancymay be the reason behind theweak electron density of the H subunit, which, if true,would imply that the enzyme structure is significantlydestabilized or flexible in the absence of the nucleo-tide. Our inability to get crystals of the apoenzyme todiffract at even medium resolution supports thishypothesis. Glycerolmoleculeswere also found in theactive-site crevice. In addition, two clear electrondensities, too high to be water molecules, were foundin the original map of each subunit. They correspondto K+ ions, with B values close to the average B valuesof the neighboring atoms. One of these ions occupies

Fig. 1. (a) Fold and secondary structure elements of tetrameric PaBADH exemplified by subunits A and B. In subunitA, the coenzyme binding domain is shown in blue, the catalytic domain is shown in green and the oligomerizationdomain is shown in magenta. The two K+ ions per subunit are shown as yellow balls in subunit A and as salmon balls insubunit B. The NADP+ molecule is represented as spheres using atom-coded colors. Helix G from subunit A and that fromsubunit B are enclosed in a red dashed ellipse. (b) Topology diagram of a subunit of PaBADH. Domains are shown usingthe same color code as in (a). The figure was prepared using CCP4mg,34 PDBsum35 and TopDraw.36



the intrasubunit site for a Na+ ion found in ALDH2,32

and the other occupies an intersubunit site, so far notdescribed in any other ALDH structure, located at thebase of the G α-helices (Fig. 1a). The G α-helices fromsubunits A and B are placed at one end of the internalcylinder, and those from subunits C and D are at theother end, closing the central cavity.The topology of PaBADH (Fig. 1b) is very similar to

the topologies of all ALDHs so far described. It differsfrom that of the cod BADH in that the PaBADHstructure does not possess the N-terminal α-helix thathas so far been found in the cod enzyme only.15

Catalytic cysteine conformation and oxidationstatus

Several published ALDH structures16,20,37–39 showthat the catalytic cysteinyl residue can adopt twoconformations: one,whichwewill call the “attacking”conformer, with a χ torsion angle of 64°, and another,which we will call the “resting” conformer, with a χtorsion angle of −64°. In each of the holo-PaBADHsubunits, only the resting conformer of C286(C302)was observed. The Sg of C286 is at less than van derWaals distance (between 3.08 and 3.20Å) from its ownamide nitrogen\which suggests some interactionbetween the atoms\and hydrogen bonded to awatermolecule that in turn is bonded to the carbonyloxygen of N153(N169), the strictly conserved aspar-agine residue that formspart of the so-calledoxyanionhole.40

In subunits C and E, C286(302) is oxidized tosulfenic acid [s-(hydroxy)cysteine,Cys-SOH] (Fig. 2a),whereas in subunits A, B, D, F and G, this residueforms a mixed disulfide with 2-mercaptoethanol [s-(thioethylhydroxy)cysteine] (Fig. 2b). Only in subunitH does C286(302) appear not to be modified (Fig. 2c).The Od of the sulfenic acid in subunit C has twopositions, with occupancy factors of 0.5 each, at anangle of 70° with respect to each other (Fig. 2a). Thisrules out a sulfenic acid in which the two oxygenatoms are at an angle of 109.08°.41 In some ALDHstructures, an extra electronic density is observedprotruding from the sulfur atom of the catalyticcysteines, suggesting the presence of a sulfenicacid.20,39 It is clear that there is no room for theNMN moiety of the nucleotide, in hydride transferposition, when the catalytic cysteine is oxidized (Fig.2d). Furthermore, there would be no room forsubstrate binding if an oxidized cysteine was in theattacking conformation.In subunits A, B and F, the sulfur atom of 2-

mercaptoethanol occupies almost the same positionas the sulfenic oxygen; in subunits D and G, twoconformers were found coincident with the twoconformers of the sulfenic acid observed in subunitC. This is the first time that a mixed disulfideinvolving the catalytic cysteine has been found in anALDH. Our finding indicates the ability of thecatalytic cysteine to form mixed disulfides withthiols after its oxidation to sulfenic acid. This couldprovide a mechanism for protection against irrever-sible oxidation.

Catalytic glutamyl residues: E252(E268), E387(E399) and E464(E476)

In some ALDHs, the conserved glutamyl residuesequivalent to the PaBADH residues E252(E268) andE387(E399) have been proposed to be the base thatdeprotonates the catalytic cysteine, either directly orthrough an intervening water molecule.9,10,40,42 In theholo-PaBADH crystal, the carboxylate oxygens ofE252(E268) and E387(E399) are at distances of 8 and5 Å, respectively, away from the sulfur of the catalyticcysteines. E387(E399) has the same conformationpresent in most of the ALDH structures so fardetermined; it is hydrogen bonded to the carbonyloxygen of the main chain of C286(C302) through awater molecule. Therefore, the PaBADH structuredoes not show how any of these glutamyl residuescould activate the catalytic cysteine. Both glutamylresidues have also been involved, in differentALDHs,in the deprotonation of thewatermolecule that attacksthe carbonyl carbon of the intermediate thioester, thusreleasing the acidic product of the reaction.9–11,43 Inaddition, the residue equivalent to PaBADH E387(E399) participates in the binding of the ribose of thenicotinamide portion of the nucleotide, bothwhen thenucleotide is in the hydride transfer conformation andin the hydrolysis conformation.17,32,44 These interac-tions are not observed in the PaBADH crystal becausethere is no electron density for theNMNportion of theNADP+ bound.The residue equivalent to PaBADH E252(E268) is

very flexible. Its side chain has at least three stablestates: One (which we will call the “inside” confor-mation) is pointing toward the catalytic cysteine in theattacking conformation, forming a hydrogen bondwith its sulfur. As in this position the side chain of theglutamic would have steric clashes with the NMNmoiety of the nucleotide when it is in the “hydridetransfer” conformation, the inside conformation hasbeen observed only in apo forms20,39,45 and in holoforms in which the nucleotide is not in thisconformation.20,32,38,45 The second is an “intermediate”conformation positioned at 7–8 Å from the catalyticcysteine and hydrogen bonded to a water molecule,which is close to the cysteine. This conformation is thatmost commonly found in ALDHs. A third conforma-tion (the “outside” conformation) can be observed inthe crystal structures of apo and holo forms of theAsian variant of ALDH2 [PDB entries 1zum (subunitsC, F, G andH) and 2onp (subunits B, C, D, E and F)], inwhich E268, the residue equivalent to E252 ofPaBADH, is hydrogen bonded to E476 (E464 inPaBADH). Both residues are at less than 3.3 Å distancefrom the Nz atom of K178 (K162 in PaBADH). In theholo-PaBADH structure, all subunits but C presentE252(E268) in the intermediate (with an occupancyfactor of 0.75) and outside (with an occupancy factor of0.25) conformations (Fig. 3a). Subunit C has theintermediate conformation only. For comparison, inFig. 3b is depicted the inside conformation of thisglutamyl residue as observed in human ALDH2.In the intermediate conformation, the Oe1 of E252

(E268) is hydrogen bonded to a water molecule,

Fig. 2. Stereoview of C286(C302) (a) oxidized to its sulfenic form as it appears in subunit C (with a double “resting”conformation) and subunit E; (b) forming a mixed disulfide with 2-mercaptoethanol [s-(thioethylhydroxy)C286(C302)] asseen in subunits A, B, D, F and G; and (c) nonmodified as seen in subunit H. The position of C286(C302) in the restingconformation relative to the two catalytic glutamyl residues [E387(E399) and E252(E268)] is also shown. (d) Stereoview ofs-(thioethylhydroxy)C286(C302) and NADP+ molecule (colored by atom). The NAD+ molecule found in cod BADH (PDBentry 1bpw, blue) is superposed to show that the derivatized C286(C302) occupies part of the space that the NMN+moietyof NAD+ occupies in cod BADH in the resting conformation. Electron density maps (2Fo−Fc) are contoured at 1σ (green).Images were generated using CCP4mg.34


which would presumably be at the appropriatedistance to perform the hydrolysis of the thioacylintermediate of the reaction if C286(C302) had theattacking conformation. TheOe2of E252(E268)makesa stronghydrogenbondwith themain-chain carbonyl

oxygen of V453(F465) (2.83 Å), which implies that thecarboxyl group of E252(E268) is protonated in thisconformation, even at pH 7.5 (the pH level of ourcrystallization medium). The proximity of the Oe2 ofE252(E268) to the Oe1 of E464(E476) (4.0 Å) suggests

Fig. 3 (legend on next page)


Fig. 3. (a) Stereoview of the PaBADH active site showing the “intermediate” and “outside” conformations of E252(E268). The surface of the cavity is shown in standard colors for different atom types (carbon atoms in white). In cylindersare depicted certain amino acids not included in the surface calculation: K162(K178) in yellow, E464(E476) in apple green,the two alternate conformations of E252(E268) in green and C286 in dark green. The glycerol bound in the active site ofPaBADH monomer A is shown in pink. (b) Stereoview of the human ALDH2 (PDB entry 1o05) active site showing E268(corresponding to PaBADH E252) in the “inside” conformation and C302 (corresponding to PaBADH C286) in the“attacking” conformation. In (a) and (b), the yellow arrows point to the binding site of the NMN+ moiety of NAD+ (notpresent in these structures). (c) Same as (a) but with a betaine aldehyde molecule (shown as magenta and red surface)docked in the active site. (d) Stereoview of the water channel connecting E464(E476) (carbon atoms as light yellowspheres) with the solvent (top left in white). The internal surface of the channel is shown in standard colors for differentatom types with white carbon and red oxygen atoms. The crystallographic water molecules are shown as dark redspheres. Images were generated using CCP4mg.34


an electrostatic interaction between their carboxylicgroups that would result in an increased pKacompared with that of E252(E268). The Oe2 of E464is making an ionic bondwith the Nz of K162(K178)—at an averagedistance of 2.9Å—and its carboxylic pKashould therefore be much lower than that of E252(E268). In the outside conformation, E252(E268) isplaced at less than 3.2 Å of K162(K178), which shouldlower the pKa of the carboxylic group. In fact, in thisconformation of E252(E268), K162(K178) is almostequidistant from E252(E268) and E464(E476), whichshould therefore have similar pKa values, and theirprotonation states could be interchanged. Thus, anefficient transfer of the proton from E252(E268) toE464(E476) could be achieved when E252(E268) is inthe outside conformation. E464(E476) can thentransfer the proton to a water molecule to which it ishydrogen bonded and that forms part of a channel ofwater molecules connecting with the surface bulk

water (Fig. 3d). Once the proton is transferred fromE464(E476) to the bulk water, the electrostatic repul-sion between a deprotonated E252(E268) and adeprotonated E464(E476) would expel the formertoward the inside and intermediate positions,where itcould deprotonate the catalytic cysteine and thehydrolytic water molecule, respectively. Therefore,the PaBADH structure suggests the existence of aproton relay system\formed by the two well-conserved residues E464(E476) and K162(K178)\that allows the deprotonation of E252(E268), asshown in Scheme 1.We found that the residues equivalent to E252

(E268), E464(E476) and K162(K178) in several ALDHcrystallographic structures have the same interactionsas those observed in PaBADH. Thus, in every ALDHstructure that has the catalytic glutamic equivalent toE252(E268) in the intermediate conformation [withthe exception of the holo-Δ1-pyrroline-5-carboxylate

Scheme 1. Mechanism for the proposed proton relay system in PaBADH. (1) A deprotonated E252(E268) in theintermediate conformation takes a proton from the hydrolytic water (or from the catalytic cysteine, not shown in thescheme). (2) The protonated E252(E268) is stabilized by hydrogen bonding to the main-chain carbonyl oxygen of V453(F465). (3) E252(E268) moves to the outside conformation and forms a hydrogen bond with E464(E476), thus allowing theproton transfer. (4) When the protonated E464 loses the proton to a water molecule of the water channel, E252(E268)moves back to the intermediate conformation (or to the inside conformation, not shown in the scheme, if the NMN+

moiety of the nucleotide is not bound).


dehydrogenase of Thermus thermophilus (TtP5CDh);PDB entries 2bhp and 2ehq] so far determined, thisresidue shows the same hydrogen bond with themain-chain carbonyl oxygen of the residue equivalentto V453(F465), implying that the glutamic is alsoprotonated in these structures.Moreover, a look at theamino acid sequences of ALDHs indicates that manyof them\86% of 546 nonredundant sequences\havethe three residues. Others have aspartyl residues at aposition equivalent to that of E464(E476) or, lessfrequently, an arginyl residue at a position equivalentto K162(K178). It therefore seems that a proton relaysystem involving the residues equivalent to E252(E268), E464(E476) and K162(K178) operates in manyALDHs. A proton relay system involving the catalyticglutamyl equivalent to E252(E268) of PaBADH andwater molecules was proposed to occur in 10-

formyltetrahydrofolate dehydrogenase (FDH),20 butthe researchers did not realize the likely participationof the aspartyl and lysyl residues that in this structureare the equivalents of E464(E476) and K162(K178) ofPaBADH. The possible role in the chemical mechan-ism of many ALDHs of the residues equivalent toE464(E476) andK162(K178) has not been realized andthus deserve further investigation. We think theseresidues could be of relevance not only because theywould participate in the proton relay mechanism thatwe are proposing and provide a sort of switch bywhich the glutamyl equivalent to E252(E268) maymove from one position to another during catalysisbut also because they help in structuring the regionthat binds the nucleotide (see below).In those ALDH structures that do not have the

acidic and basic residues equivalent to E464(E476)


and K162(K178), such as the TtP5CDh structure, thecatalytic glutamyl equivalent to E252(E268) ofPaBADH is hydrogen bonded to a water moleculethat forms part of a channel of water moleculesconnected to the protein surface and can thereforerelease the proton directly to the bulk water withoutanother intervening glutamyl residue.It is important to point out that when the enzyme

has the aldehyde and the nucleotide in the hydrideconformation bound, the protonated glutamylequivalent to E252(E268) in the intermediate con-formation cannot transfer its proton directly to awater molecule because the amide group of theNMN+ occupies the position of the hydrolyticwater,20,32 and betaine aldehyde appears to blockthe entrance of water from the aldehyde bindingsubsite (Fig. 3c). Once the hydride transfer has takenplace and the resulting NMNH moiety of thenucleotide has moved into the hydrolysis position,the hydrolyticwatermolecule can enter the active site,but the access of other water molecules from thealdehyde binding subsite to the proximity of E252(E268) should still be blocked by the reactionintermediate covalently linked to the catalyticcysteine. Thus, the only way in which the residueequivalent to E252(E268) can lose the proton takenfrom either the catalytic cysteine or the hydrolyticwater is through the proton relay system. There aretherefore clear reasons for the change of conformationof this glutamyl during the catalytic cycle. In theinside conformation, it can deprotonate the catalyticthiol but would interfere with the binding of thenucleotide and would be too close to the thioacylintermediate to allow the positioning of the hydrolyticwater, as pointed out by others.20,32,39 One canspeculate that the inside conformation is not stableafter the glutamyl residue becomes protonated byabstraction of the proton from the catalytic cysteine,so it moves to a more stable intermediate conforma-tion. Additionally, because in the inside conformationthe carboxyl group is in the same position as theamide of the NMN+ in the hydride transferconformation,20,39 the binding of the nucleotideshould help the residue equivalent to E252(E268)move away from the inside conformation to the

Fig. 4. Stereoview of the glycerol molecule bound in the PaBmap is contoured at 1σ and was generated using CCP4mg.34

intermediate conformation. In the intermediate con-formation, it cannot directly deprotonate the catalyticcysteine but can deprotonate the hydrolytic watermolecule. In the outside conformation, it can releasethe proton, taken either from the catalytic cysteine orfrom the hydrolytic water, via the proton relaysystem. As a deprotonated glutamyl is not stable inthis conformation, it has to move back to the insideor intermediate conformation, depending on thepresence and conformation of the nucleotide. Theactive site of several ALDHs appears to haveevolved to accommodate and stabilize these threeconformations.

Glycerol binding

Electron density that could be fitted with glycerolmolecules was found in the PaBADH structure. Someof these molecules were at the subunit surface orbetween subunits, but at least one of themwas in eachof the active sites. In the latter position, glycerolbinding does not appear to be a mere artifact ofcrystallization and rather appears to mimic thebinding of the aldehyde substrate. In subunits Aand B, the O1 of the glycerol molecule is makinghydrogen bondswith theNd2 ofN153(N169) (2.85 Å)and, in subunit A, with the amide nitrogen of C286(C302) (3.32 Å) as well (Fig. 4). Both amide nitrogenatoms constitute the oxyanion hole that stabilizes thethiohemiacetal and thioacyl intermediates by formingthe hydrogen bonds that in our structure form thehydroxyl oxygen of glycerol.40 The glycerolmoleculesin subunits A and B therefore mimic the thiohemia-cetal intermediate,which has so far not been observedin any crystal. However, the first tetrahedral carbon ofthe glycerolmolecule is not exactly in the position thatit would be in if the aldehyde substrate forms acovalent bond with the cysteine sulfur, thus allowingtransfer of the hydride to the nucleotide. The samehydrogen bonds with the amide nitrogen atoms thatform the oxyanion hole have been observed with thecarbonyl oxygen of glyceraldehyde-3-phosphatebound to the active site of the catalytic C→S mutantof the nonphosphorylating glyceraldehyde-3-phos-phate dehydrogenase (GAPN) of Streptococcus

ADH active site of subunit A. The 2Fo−Fc electron densityDistances are in angstrom.


mutans,46 the thioacyl intermediate in the GAPN of S.mutans37 or the Oe1 of the of the carboxylic group ofthe glutamate bound to the active site of TtP5CDh.39

In the other PaBADH subunits, the glycerolmolecules are displaced between 1.2 and 3.2 Åwith respect to their position in subunits A and B, andtheir Oe1 is hydrogen bonded to a water moleculethat links the glycerol oxygenwith the amide nitrogengroups of C286(C302) and the Nd2 of N153(N169)(data not shown). These glycerol molecules suggestthe pathway that the aldehyde substrate follows in itsentrance to the active site previous to its finalproductive binding. Glycerol used both in the crystal-lization medium and as a cryoprotectant can also beobserved in the active site of the FDH crystals, but inthis case, the glycerol molecules do not make thehydrogen bonds with the nitrogen of the two amidegroups of the oxyanion hole.

Binding of NADP+

Only the ADP moiety of the NADP+ molecule wasobserved in the PaBADH structure. The lack ofelectronic density for the NMN+ moiety in thePaBADH crystal seems to be the result of occupancyof its cavity by the oxidized catalytic cysteine (Fig. 2d).Despite this, it is interesting that in four of the eightPaBADH subunits (A, F, G and H) that form theasymmetric unit, the pyrophosphate is bent as if theNMN+ portion of the nucleotidewas oriented towardthe active site,while in the other four (B, C,D andE), itis bent as expected for a nonbound NMN+ moiety.PaBADH has the glutamyl residue [E179(E195)]

that has been previously thought to be incompatiblewith the binding of the 2′-phosphate of NADP+ andtherefore characteristic of NAD+-dependentALDHs.14 In the case of PaBADH, the side chain of

Fig. 5. Stereoview of the binding site of the 2′-phosphate(E53) are shown using stick representations. The atoms and thePyMOL.47 Distances are in angstrom.

this glutamyl is kept away from the binding site ofthe 2′-phosphate by a strong ionic interactionbetween the Oe1 of E179(E195) and the Nh1 of R40(E53) (with an average distance of 2.77 Å) (Fig. 5).Some other ALDHs that bindNADP+ with similar orhigher affinity than NAD+ have a nonchargedresidue at a position equivalent to that of E179(E195),16–18,20 whereas this glutamyl residue forms ahydrogen bond with the 2′-phosphate in theTtP5CDh.19 Thus, in spite of a high level of similarityin sequence and structure, different ALDHs usedifferent solutions to the problem of binding the 2′-phosphate group of NADP+.

Binding of K+ ions

Two clear electron density peaks, with valueshigher than 8σ in a 2Fo−Fc map, were observed ineach PaBADH subunit: one inside each subunit andthe other in the interface between subunits in adimeric unit (Fig. 6). We conclude that the cationbound by PaBADH is K+ partially because of theabsence of Na+ but the presence of 150 mMK+ in ourcrystallizationmediumand fundamentally because ofthe strong electron density observed, more suitablefor K+ (19 electrons) than for Na+ (11 electrons). Thefirst peak corresponds to aK+ ion hexa-coordinated toOg1 of T26(T39), O of I27(V40), Od1 and O of D93(D109), O of V180(N196) and a water molecule (Fig.6). This cation binding site is the same as the onedescribed for a Na+ ion in ALDH2,32 where it wassuggested that it contributes to the optimal active-siteconformation for binding the nucleotide. This role hasnot yet been experimentally confirmed, however.A revision of the structures so far deposited in the

PDB revealed that many of them have a watermolecule in the same position as the intrasubunit

of NADP+ in PaBADH. The NADP+, E179(E195) and R40surface are colored by atom. Images were generated using

Fig. 6. Stereoview of the intrasubunit binding site of a K+ ion in a 2Fo−Fc electron density map contoured at 1σ (green)and 5σ (blue). The image was generated using CCP4mg.34


cation found in ALDH2 and PaBADH. Because it isdifficult to discriminate betweenwater andNa+ by X-ray crystallography (with 11 and 10 electrons,respectively), it could be that the electron densityassigned to awatermolecule in those structures in factcorresponds to the monovalent cation present in thecrystallization media. Particularly, this seems to betrue in those proteins crystallized in the presence ofammonium sulfate or a sodium salt\such as lactal-dehyde dehydrogenase from E. coli, FDH, TtP5CDhand GAPN from Thermoproteus tenax\that show fiveor six possible hydrogen bonds between the proteinand the water molecule inside this cavity, whichsuggests that these hydrogen bonds are indeed thecation coordination bonds. In other ALDH crystals,there is clearly a water molecule, given that it makesfour hydrogen bonds and that the crystals wereobtained in a very low monovalent cation concentra-tion (for instance, cod BADH). Therefore, the presenceof this intrasubunit cavity and its ability to bind amonovalent cation appear to be common features ofmany ALDHs.The other electronic density peak corresponds to a

second K+ ion located in the subunit interface withineach dimer. There are therefore two sites insymmetrical positions in each of the interfaces of Awith B (Fig. 1a), C with D, E with F and G with H.The intersubunit K+ ions are coordinated by atomsbelonging to residues K457(K469) (O) and G460(G472) (O) of a loop (residues 455–467 in PaBADH,corresponding to residues 467–479 in ALDH2) ofone subunit, to L246(L262) (O) of the neighboringsubunit and to three water molecules (Fig. 7a).Additionally, the C-terminal end of the G α-helix ofthe neighboring subunit, and for extension theresidual negative charge of its dipole, is at 7 Å ofthe intersubunit K+. In subunit C, the Oe2 of residueE248(K264) substitutes one of these water moleculesin coordinating the cation (Fig. 7b). The K+ ion is atthe bottom of a tunnel blocking the connectionbetween the surface and the internal cavity of theprotein (Fig. 7c). Because the intersubunit K+ ion isinvolved in a net of interactions that stabilize thedimer, it should play an important role in thestabilization of the native PaBADH tetrameric

structure. This is consistent with the previousfinding that PaBADH dissociates into subunits in theabsence of K+ ions in a buffer of low ionic strength.22

E248(K264) seems to be a necessary counterchargeto the K+ cation in PaBADH. No other ALDH ofknown three-dimensional structure has an acidicresidue in the position equivalent toE248 of PaBADH.In ALDH2 and ALDH1, the residue equivalent toE248 is an arginine, R264 in ALDH2 and ALDH1numbering. Looking at several electronic densitymaps of these enzymes, we found that the Nh1 orNh2 of the guanidinium group occupies almost thesame position as the K+ ion in PaBADH, makinghydrogen bonds with residues belonging to twoneighboring subunits. This may explain why in theseALDHs it has not been observed that the absence ofcations affects the structure of the activity of theenzyme. It is interesting that one of the hydrogenbonds of R264 is with E487 (ALDH2 numbering), theresidue whose natural mutation to a lysine causesalcohol intolerance in Asians.46 This Asian variant ofALDH2 is a very inefficient enzyme, mostly due to itsvery low affinity for the nucleotide46 caused by a highdegree of disorder in the regionof helixGand the loop466–479 (corresponding to the loop 454–467 inPaBADH).48 Interestingly, in the Asian variant ofALDH2, the side chain of R264 is pointing away fromthe center of the cation binding cavity, which has noother cation bound. The only structured crystal of theAsian variant of ALDH2 is a holo form (PDB entry2onp) in which a guanidinium ion from the crystal-lization medium occupies a position almost equiva-lent to that of the intersubunit K+ ion in PaBADH.There may be a correlation between the order in thisregion and the simultaneous binding of the inter-subunit monovalent cation and the nucleotide. Thismay be the reason behind the synergic effects of thenucleotide and K+ ions in reactivating the PaBADHpreviously inactivated by incubation in a bufferdevoid of monovalent cations.21 The exception maybe the retinal dehydrogenase, which shows disorderin this region in spite of having the arginyl residueNAD+ bound and crystallized in the presence of ahigh concentration of ammonium sulfate. This enzymeappears to require the binding of the proper substrate

Fig. 7. (a) Stereoview in a 2Fo−Fc electron density map of the binding site between subunit A (magenta worm) andsubunit B (blue worm) of a K+ ion, contoured at 1σ (green). In subunit A, the loop formed by residues 455–464 isrepresented as a red worm. (b) Stereoview in a 2Fo−Fc electron density map of the binding site between subunit C (purpleworm) and subunit D (cyan worm) of a K+ ion, contoured at 1σ (green). Here, E248(K264) provides a coordination bond,substituting for a water molecule. (c) Stereoview of the channel communicating the protein surface with the K+ ionbinding site between subunit A (depicted as a magenta worm) and subunit B (depicted as a blue worm). The channelsurface is colored representing its electrostatic potential (+0.5 V in blue; 0 V in white; −0.5 V in red). K+ ions arerepresented as yellow balls; water molecules, as red balls. The figure was prepared with CCP4mg.34


to become ordered, and this constitutes the basis for itssubstrate specificity.49 The YcdW ALDH and theTtP5CDh also have the arginine equivalent to E248 ofPaBADH, but its side chain is outside the cationbinding cavity, and still their structures show no

disorder. The electron density map of these crystalsindicates that it is likely that they do have a cationbound inside this cavity. As mentioned before for theintrasubunit cation binding cavity, the five crystals ofTtP5CDh showawatermoleculemaking six hydrogen


bonds. This water is most likely a Na+ from thecrystallization medium. The electron density maps ofother ALDHs that have a proline (ALDH from Vibrioharveyi) or a lysine (FDH) in the position equivalent toE248 of PaBADH show water molecules inside theintersubunit cation binding cavity, but, for the reasonsmentioned above, we think these water molecules areindeed Na+ cations. Cod BADH has a histidine and awater molecule (PDB entry 1a4s), but it was crystal-lized in a low cation concentration medium. The onlydimeric structure so far known of an ALDH, that ofALDH3 (PDB entry 1ad3), also has awatermolecule inthis cavity.Therefore, it seems that this intersubunit cation

binding cavity is also a general feature of theALDHs,at least of the tetrameric enzymes. The binding of apositive charge into this cavity appears to stabilize aloop that is essential for both catalysis (through therole proposed for E464 in proton release) andcoenzyme binding (through the stabilization ofhelix G), as well as for tetramerization, as observedexperimentally in PaBADH.21,22

Materials and Methods

Expression and purification of PaBADH

Plasmid pCALbetB containing the full sequence of thegene betB that encodes PaBADH was used for theexpression of the enzyme in E. coli cells as previouslydescribed.4 The recombinant enzyme was purified tohomogeneity by a two-step procedure previouslydescribed.50 Protein concentrations were determined bythe Coomassie G dye binding technique of Bradford51

using bovine serum albumin as a standard or spectro-photometrically using a molar absorptivity at 280 nm of52,060M−1 cm−1, deduced from the amino acid sequence bythe method of Gill and von Hippel.52

Crystallization and data collection

BADHwas concentrated to 20mg/ml in anAmiconUltra30 (Millipore), dialyzed exhaustively against 10 mM potas-sium phosphate buffer, pH 6.9, containing 0.2 mM ethyle-nediaminetetraacetic acid, 300 mM KCl, 1% (v:v) glyceroland 20 mM 2-mercaptoethanol and crystallized underaerobic conditions using the hanging-drop vapor-diffusionmethod. Crystals were grown in solution 41 of the crystalscreen Cryo of Hampton Research (Aliso Viejo, CA, USA),which contained 85 mM Hepes–NaOH, pH 7.5, 8.5% (v:v)isopropanol, 17% (w:v) polyethylene glycol 4000 and 15%(v:v) glycerol. Crystals of PaBADH with the coenzyme(1 mM NADP+, sodium salt) were obtained by adding thiscompound to the mix prior to crystallization. The dropswere formed with 1 μl of protein plus 1 μl of precipitantsolution. Trays containing the holoenzyme crystals wereincubated at 18 °C. BADH crystals appeared after 3 daysand were grown for about 2 weeks until they reached theirmaximum dimensions. Crystals grew as large thin plateswith irregular shapes, square or rectangular. The crystalswere cooled under nitrogen stream at 100 K during datacollection. Synchrotron data were collected at the NationalSynchrotron Light Source (Upton, NY, USA) on beam lineX6a. The data were auto-indexed and integrated using

MOSFLM53 and then scaled and truncated with programsfrom the CCP4 suite.54

Structure solution and refinement

PaBADH phases were determined by molecular repla-cement with MOLREP55 using the coordinates of ourPaBADH computational homology model50 as a startingmodel and locked self-rotation functions. In brief, to get aconvincing solution, the noncrystallographic symmetrieswere first detected and the rotation and translationsearches were then constrained around the noncrystallo-graphic symmetries' peaks. We carried out alternatingcycles of automatic refinement and manual adjustmentsusing annealing and minimizing individual B values withthe standard protocols of CNS version 1.0.56 The programCoot57 was used to analyze the electron density maps(2Fo−Fc and Fo−Fc) in order to build some parts of thePaBADH model, particularly the subunit H, and formanual refinement. Parameters for proper stereochemicalrefinement of the bound NADP+ were copied from theHIC-UP (Hetero Compound Information Center) site.58

Water molecules were automatically localized and gener-ated by Coot.57 A TLS refinement under REFMAC 5 wasperformed in order to analyze the occurrence of a probablepseudo-rigid-body displacement in PaBADH crystals.59

After several trials, the division of the molecule in thetraditional ALDH domains produced a significantdecrease in the residual refinement values (4%, both in Rand Rfree). Structural alignments with cod liver BADH andother ALDHs were done with the Coot57 and CCP4mgprograms.34

PDB accession number

The coordinates and the structure factors for the structureof PaBADH have been deposited with the ResearchCollaboratory for Structural Bioinformatics PDB withaccession code 2ve5.

Acknowledgements

This research was financially supported by Direc-ción General de Asuntos del Personal Académico dela Universidad Nacional Autónoma de México(IN228106) and Consejo Nacional de Ciencia yTecnología de México (50581). We thank Dr. AdelaRodríguez-Romero from the Laboratorio Universi-tario de Estructura de Proteínas, UniversidadNacional Autónoma de México, and Instituto deFísica de San Carlos, Universidade de São Paulo,Brazil, for the initial BADH data sets and the staff atbeam line X6a of the National Synchrotron LightSource for the final BADH data collection sets. Wealso thank Sonia Rojas-Trejo for technical assistancewith the crystallization assays.

References

1. Rothschild, H. A. & Guzman Barron, E. S. (1954). Theoxidation of betaine aldehyde by betaine aldehydedehydrogenase. J. Biol. Chem. 209, 511–523.


2. Nagasawa, T., Kawabata, Y., Tani, Y. & Ogata, K.(1976). Purification and characterization of betainealdehyde dehydrogenase from Pseudomonas aeruginosaA-16. Agric. Biol. Chem. 40, 1743–1749.

3. Hanson, A. D., May, A. M., Grumet, R., Bode, J.,Jamieson, G. C. & Rhodes, D. (1985). Betaine synthesisin chenopods: localization in chloroplasts. Proc. NatlAcad. Sci. USA, 82, 3678–3682.

4. Velasco-García, R., Villalobos, M. A., Ramírez-Romero, M. A., Mújica-Jiménez, C., Iturriaga, G. &Muñoz-Clares, R. A. (2006). Betaine aldehyde dehy-drogenase from Pseudomonas aeruginosa: cloning, over-expression in Escherichia coli, and regulation bycholine and salt. Arch. Microbiol. 185, 14–22.

5. Diab, F., Bernard, T., Bazire, A., Haras, D., Blanco, C. &Jebbar, M. (2006). Succinate-mediated cataboliterepression control on the production of glycinebetaine catabolic enzymes in Pseudomonas aeruginosaPAO1 under low and elevated salinities. Microbiology,152, 1395–1406.

6. Sage, A. E., Vasil, A. I. & Vasil, M. L. (1997). Molecularcharacterization of mutants affected in the osmoprotec-tant-dependent induction of phospholipase C in Pseu-domonas aeruginosa PAO1.Mol. Microbiol. 23, 43–56.

7. Velasco-García, R., Chacon-Aguilar, V. M., Hervert-Hernandez, D. & Muñoz-Clares, R. A. (2003). Inacti-vation of betaine aldehyde dehydrogenase fromPseudomonas aeruginosa and Amaranthus hypochondria-cus L. leaves by disulfiram. Chem. Biol. Interact. 143–144, 149–158.

8. Velasco-García, R., Zaldívar-Machorro, V. J., Mújica-Jiménez, C., González-Segura, L. &Muñoz-Clares, R. A.(2006). Disulfiram irreversibly aggregates betaine alde-hyde dehydrogenase—a potential target for antimicro-bial agents against Pseudomonas aeruginosa. Biochem.Biophys. Res. Commun. 341, 408–415.

9. Wang, X.&Weiner,H. (1995). Involvement of glutamate268 in the active site of human livermitochondrial (class2) aldehyde dehydrogenase as probed by site-directedmutagenesis. Biochemistry, 34, 237–243.

10. Hempel, J., Perozich, J., Chapman, T., Rose, J., Boesch,J. S., Liu, Z. J. et al. (1999). Aldehyde dehydrogenasecatalytic mechanism. A proposal. Adv. Exp. Med. Biol.436, 53–59.

11. Mann, C. J. & Weiner, H. (1999). Differences in theroles of conserved glutamic acid residues in the activesite of human class 3 and class 2 aldehyde dehydro-genases. Protein Sci. 8, 1922–1929.

12. Velasco-García, R., González-Segura, L. & Muñoz-Clares, R. A. (2000). Steady-state kinetic mechanism ofthe NADP+- and NAD+-dependent reactions cata-lysed by betaine aldehyde dehydrogenase fromPseudomonas aeruginosa. Biochem. J. 352, 675–683.

13. Julián-Sanchez, A., Riveros-Rosas, H., Martínez-Cas-tilla, L. P., Velasco-García, R. & Muñoz-Clares, R. A.(2007). Phylogenetic and structural relationships ofthe betaine aldehyde dehydrogenases. In Enzymologyand Molecular Biology of Carbonyl Metabolism 13(Weiner, H., Maser, E., Lindahl, R. & Plapp, B., eds),pp. 64–76, Purdue University Press, West Lafayette,IN.

14. Perozich, J., Kuo, I., Wang, B. C., Boesch, J. S., Lindahl,R. & Hempel, J. (2000). Shifting the NAD/NADPpreference in class 3 aldehyde dehydrogenase. Eur. J.Biochem. 267, 6197–6203.

15. Johansson, K., El-Ahmad, M., Ramaswamy, S.,Hjelmqvist, L., Jörnvall, H. & Eklund, H. (1998).Structure of betaine aldehyde dehydrogenase at 2.1 Åresolution. Protein Sci. 7, 2106–2117.

16. Cobessi, D., Tête-Favier, F., Marchal, S., Azza, S.,Branlant, G. & Aubry, A. (1999). Apo and holo crystalstructures of an NADP-dependent aldehyde dehy-drogenase from Streptococcus mutans. J. Mol. Biol. 290,161–173.

17. Ahvazi, B., Coulombe, R., Delarge, M., Vedadi, M.,Zhang, L., Meighen, E. & Vrielink, A. (2000). Crystalstructure of the NADP+-dependent aldehyde dehy-drogenase from Vibrio harveyi: structural implicationsfor cofactor specificity and affinity. Biochem. J. 349,853–861.

18. Pohl, E., Brunner, N., Wilmanns, M. & Hensel, R.(2002). The crystal structure of the allosteric non-phosphorylating glyceraldehyde-3-phosphate dehy-drogenase from the hyperthermophilic archaeumThermoproteus tenax. J. Biol. Chem. 277, 19938–19945.

19. Inagaki, E., Ohshima, N., Sakamoto, K., Babayeva,N. D., Kato, H., Yokoyamaa, S. & Tahirov, T. H.(2007). New insights into the binding mode ofcoenzymes: structure of Thermus thermophilus Δ1-pyrroline-5-carboxylate dehydrogenase complexedwith NADP+. Acta Crystallogr., Sect. F: Struct. Biol.Cryst. Commun. 63, 462–465.

20. Tsybovsky, Y., Donato, H., Krupenko, N. I., Davies, C.& Krupenko, S. A. (2007). Crystal structures of thecarboxyl terminal domain of rat 10-formyltetrahydro-folate dehydrogenase: implications for the catalyticmechanism of aldehyde dehydrogenases. Biochemistry,46, 2917–2929.

21. Velasco-García, R., Mújica-Jiménez, C., Mendoza-Her-nández, G. & Muñoz-Clares, R. A. (1999). Rapidpurification and properties of betaine aldehyde dehy-drogenase from Pseudomonas aeruginosa. J. Bacteriol. 181,1292–1300.

22. Valenzuela-Soto, E. M., Velasco-García, R., Mújica-Jimé-nez, C., Gaviria-González, L. & Muñoz-Clares, R. A.(2003). Monovalent cations requirements for the stabilityof betaine aldehyde dehydrogenase from Pseudomonasaeruginosa, porcine kidney and amaranth leaves. Chem.Biol. Interact. 143–144, 139–148.

23. Garza-Ramos, G., Carrillo-Nava, E., Costas, M.,Mújica-Jiménez, C., Velasco-García, R. & Muñoz-Clares, R. (2007). Effects of monovalent cations onthe structure and thermal stability of betaine alde-hyde dehydrogenase from Pseudomonas aeruginosa. InEnzymology and Molecular Biology of Carbonyl Metabo-lism 13 (Weiner, H., Maser, E., Lindahl, R. & Plapp,B., eds), pp. 83–90, Purdue University Press, WestLafayette, IN.

24. Falkenberg, P. & Strom, A. R. (1990). Purification andcharacterization of osmoregulatory betaine aldehydedehydrogenase from Escherichia coli. Biochim. Biophys.Acta, 1034, 253–259.

25. Valenzuela-Soto, E. M. & Muñoz-Clares, R. A. (1994).Purification and properties of betaine aldehydedehydrogenase extracted from detached leaves ofAmaranthus hypochondriacus L. subjected to waterdeficit. J. Plant Physiol. 143, 145–152.

26. Boch, J., Nau-Wagner, G., Kneip, S. & Bremer, F.(1997). Glycine betaine aldehyde dehydrogenase fromBacillus subtilis: characterization of an enzymerequired for the synthesis of the osmoprotectantglycine betaine. Arch. Microbiol. 168, 282–289.

27. Guzmán-Partida, A. M. & Valenzuela-Soto, E. M.(1998). Porcine kidney betaine aldehyde dehydrogen-ase: purification and properties. Comp. Biochem.Physiol. 119, 485–491.

28. Dragolovich, J. & Pierce, S. K. (1994). Characterizationof partially purified betaine aldehyde dehydrogenase


from horseshoe crab (Limulus polyphemus) cardiacmitochondria. J. Exp. Zool. 270, 417–425.

29. Von Tigerstrom, R. G. & Razell, W. E. (1968). Aldehydedehydrogenase: II. Physical and molecular propertiesof the enzyme from Pseudomonas aeruginosa. J. Biol.Chem. 243, 6495–6503.

30. Sorger, G. J. & Evans, H. J. (1966). Effects of univalentcations on the properties of yeast NAD+ acetaldehydedehydrogenase. Biochim. Biophys. Acta, 118, 1–8.

31. Wang, X., Mann, C. J., Bai, Y., Ni, L. & Weiner, H.(1998). Molecular cloning, characterization, andpotential roles of cytosolic and mitochondrial alde-hyde dehydrogenases in ethanol metabolism inSaccharomyces cerevisiae. J. Bacteriol. 180, 822–830.

32. Perez-Miller, S. J. & Hurley, T. D. (2003). Coenzymeisomerization is integral to catalysis in aldehydedehydrogenase. Biochemistry, 42, 7100–7109.

33. Winn, M. D., Isupov, M. N. & Murshudov, G. N.(2001). Use of TLS parameters to model anisotropicdisplacements in macromolecular refinement. ActaCrystallogr., Sect. D: Biol. Crystallogr. 57, 122–133.

34. Potterton, L., McNicholas, S., Krissinel, E., Gruber, J.,Cowtan, K., Emsley, P. et al. (2004). Developments inthe CCP4 molecular-graphics project. Acta Crystallogr.,Sect. D: Biol. Crystallogr. 60, 2288–2294.

35. Laskowski, R. A., Chistyakov, V. V. & Thornton, J. M.(2005). PDBsum more: new summaries and analysesof the known 3D structures of proteins and nucleicacids. Nucleic Acids Res. 33, D266–D268.

36. Bond, C. S. (2003). TopDraw: a sketchpad for proteinstructure topology cartoons. Bioinformatics, 19, 311–312.

37. D'Ambrosio, K., Pailot, A., Talfournier, F., Didierjean,C., Benedetti, E., Aubry, A. et al. (2006). The firstcrystal structure of a thioacylenzyme intermediate inthe ALDH family: new coenzyme conformation andrelevance to catalysis. Biochemistry, 45, 2978–2986.

38. Gruez, A., Roig-Zamboni, V., Grisel, S., Salomoni, A.,Valencia, C., Campanacci, V. et al. (2004). Crystalstructure and kinetics identify Escherichia coli YdcWgene product as a medium-chain aldehyde dehydro-genase. J. Mol. Biol. 343, 29–41.

39. Inagaki, E., Ohshima, N., Takahashi, H., Kuroishi, C.,Yokoyama, S. & Tahirov, T. H. (2006). Crystal structureof Thermus thermophilus Δ1-pyrroline-5-carboxylatedehydrogenase. J. Mol. Biol. 362, 490–501.

40. Steintmetz, C. G., Xie, P., Weiner, H. & Hurley, T. D.(1997). Structure of mitochondrial aldehyde dehydro-genase: the genetic component of ethanol aversion.Structure, 5, 701–711.

41. Canet-Avilés, R. M., Wilson, M. A., Miller, D. W.,Ahmad, R., McLendon, C., Bandyopadhyay, S.,Baptista, M. J. et al. (2004). The Parkinson's diseaseprotein DJ-1 is neuroprotective due to cysteine-sulfinic acid-driven mitochondrial localization. Proc.Natl Acad. Sci. USA, 101, 9103–9108.

42. Vedadi, M. & Meighen, E. (1997). Critical glutamicacid residues affecting the mechanism and nucleotidespecificity of Vibrio harveyi aldehyde dehydrogenase.Eur. J. Biochem. 246, 698–704.

43. Marchal, S., Rahuel-Clermont, S. & Branlant, G.(2000). Role of glutamate-268 in the catalytic mechan-ism of nonphosphorylating glyceraldehyde-3-phos-phate dehydrogenase from Streptococcus mutans.Biochemistry, 39, 3327–3335.

44. Ni, L., Sheikh, S. & Weiner, H. (1997). Involvement ofglutamate 399 and lysine 192 in the mechanism of

human liver mitochondrial aldehyde dehydrogenase.J. Biol. Chem. 272, 18823–18826.

45. Cobessi, D., Tête-Favier, F., Marchal, S., Branlant, G. &Aubry, A. (2000). Structural and biochemical investiga-tions of the catalytic mechanism of an NADP-depen-dent aldehyde dehydrogenase from Streptococcusmutans. J. Mol. Biol. 300, 141–152.

46. Farrés, J., Wang, X., Takahashi, K., Cunningham, S. J.,Wang, T. T. & Weiner, H. (1994). Effects of changingglutamate 487 to lysine in rat and human livermitochondrial aldehyde dehydrogenase. A model tostudy human (oriental type) class 2 aldehyde dehy-drogenase. J. Biol. Chem. 269, 13854–13860.

47. DeLano, W. L. (2002). The PyMOL Molecular Gra-phics System on the World Wide Web. http://www.pymol.org/

48. Larson, H. N., Weiner, H. & Hurley, T. D. (2005).Disruption of the coenzyme binding site and dimerinterface revealed in the crystal structure of mitochon-drial aldehyde dehydrogenase “Asian” variant. J. Biol.Chem. 280, 30550–30556.

49. Bordelon, T., Montegudo, S. K., Pakhomova, S.,Oldham, M. L. & Newcomer, M. E. (2004). A disorderto order transition accompanies catalysis in retinalde-hyde dehydrogenase type II. J. Biol. Chem. 279,43085–43091.

50. González-Segura, L., Velasco-García, R., Rudiño-Piñera, E., Mújica-Jiménez, C. & Muñoz-Clares, R. A.(2005). Site-directed mutagenesis and homologymodeling indicate an important role of cysteine 439in the stability of betaine aldehyde dehydrogenasefrom Pseudomonas aeruginosa. Biochimie, 87, 1056–1064.

51. Bradford, M. M. (1976). A rapid and sensitive methodfor the quantitation of microgram quantities of proteinutilizing the principle of protein-dye binding. Anal.Biochem. 72, 248–256.

52. Gill, S. C. & von Hippel, P. H. (1989). Calculation ofprotein extinction coefficients from amino acidsequence data. Anal. Biochem. 182, 319–326.

53. Leslie, A. G. W. (1992). Recent changes to theMOSFLM package for processing film and imageplate data. Jt. CCP4+ESF-EAMCB Newslett. ProteinCrystallogr. 26, 22–33.

54. Collaborative Computational Project, Number 4.(1994). The CCP4 suite: programs for protein crystal-lography. Acta Crystallogr., Sect. D: Biol. Crystallogr. 50,760–763.

55. Vagin, A. & Teplyakov, A. (1997). MOLREP: anautomated program for molecular replacement. J.Appl. Crystallogr. 30, 1022–1025.

56. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano,W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).Crystallography & NMR System (CNS): a newsoftware suite for macromolecular structure deter-mination. Acta Crystallogr., Sect. D: Biol. Crystallogr.54, 905–921.

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

58. Kleywegt, G. J. & Jones, T. A. (1998). Databases inprotein crystallography. Acta Crystallogr., Sect. D: Biol.Crystallogr. 54, 1119–1131.

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

the crystal structure of a ternary complex of betaine ... · molecular y bioprocesos, instituto de...

Documents