www.elsevier.com/locate/yabbi

ABBArchives of Biochemistry and Biophysics 457 (2007) 123–133

Kinetic and chemical mechanisms of shikimate dehydrogenasefrom Mycobacterium tuberculosis

Isabel O. Fonseca a,b, Rafael G. Silva a,b, Claudia L. Fernandes c,Osmar N. de Souza c, Luiz A. Basso a,*, Diogenes S. Santos a,*

a Centro de Pequisas em Biologia Molecular e Funcional, Pontifıcia Universidade Catolica do Rio Grande do Sul (PUCRS),

Porto Alegre, RS 90619-900, Brazilb Programa de Pos-Graduacao em Ciencias Biologicas: Bioquımica, Universidade Federal do Rio Grande do Sul (UFRGS), Brazil

c Laboratorio de Bioinformatica, Modelagem e Simulacao de Biossistemas – LABIO, Faculdade de Informatica, PUCRS, Porto Alegre, RS 90619-900, Brazil

Received 8 September 2006, and in revised form 13 November 2006Available online 21 November 2006

Abstract

Mycobacterium tuberculosis shikimate dehydrogenase (MtbSD) catalyzes the fourth reaction in the shikimate pathway, the NADPH-dependent reduction of 3-dehydroshikimate. To gather information on the kinetic mechanism, initial velocity patterns, product inhibi-tion, and primary deuterium kinetic isotope effect studies were performed and the results suggested a steady-state ordered bi-bi kineticmechanism. The magnitudes of both primary and solvent kinetic isotope effects indicated that the hydride transferred from NADPH andprotons transferred from the solvent in the catalytic cycle are not significantly rate limiting in the overall reaction. Proton inventory anal-ysis indicates that one proton gives rise to solvent isotope effects. Multiple isotope effect studies indicate that both hydride and protontransfers are concerted. The pH profiles revealed that acid/base chemistry takes place in catalysis and substrate binding. The MtbSD 3Dmodel was obtained in silico by homology modeling. Kinetic and chemical mechanisms for MtbSD are proposed on the basis of exper-imental data.� 2006 Elsevier Inc. All rights reserved.

Keywords: Shikimate dehydrogenase; Tuberculosis; Drug target; 3-Dehydroshikimate; Shikimate; Kinetic mechanism; Chemical mechanism; Mycobac-

terium tuberculosis

Tuberculosis (TB)1 is a pandemic, which even today,remains a major global health concern. Its causative agent,Mycobacterium tuberculosis, is one of the most prolificinfectious agents affecting humans. A third of the world’spopulation is thought to host M. tuberculosis and approx-imately 30 million people have died from the disease in the

0003-9861/$ - see front matter � 2006 Elsevier Inc. All rights reserved.

doi:10.1016/j.abb.2006.11.015

* Corresponding authors. Fax: +55 51 33203629.E-mail addresses: luiz.basso@pucrs.br (L.A. Basso), diogenes@

pucrs.br (D.S. Santos).1 Abbreviations used: DHS, 3-dehydroshikimate; MDR-TB, multidrug-

resistant tuberculosis; MtbSD, Mycobacterium tuberculosis shikimatedehydrogenase; NADP+, oxidized b-nicotinamide adenine dinucleotidephosphate; NADPH, reduced b-nicotinamide adenine dinucleotide phos-phate; NADPD, deuterated b-nicotinamide adenine dinucleotide phos-phate; SD, shikimate dehydrogenase; SHK, D-shikimate; TB,tuberculosis; XDR-TB, extensively drug-resistant tuberculosis.

past decade [1]. The treatment of multidrug-resistant TB(MDR-TB), defined as resistant to at least isoniazid andrifampicin [2], requires the administration of second-linedrugs that are more toxic and less effective, and are givenfor at least three times as long as, and 100 times as expen-sive as basic chemotherapy regimens [3]. More recently, asurvey of the frequency and distribution of extensivelydrug-resistant (XDR) TB cases, which are defined as casesin persons with TB whose isolates were resistant to isonia-zid and rifampicin and at least three of the six main classesof second-line drugs, showed that during 2000–2004, of17,690 TB isolates, 20% were MDR and 2% were XDR,and that XDR-TB has a wide geographic distribution [4].New antimycobacterial agents are thus needed to improvethe treatment of MDR- and XDR-TB, and to providemore effective treatment of latent tuberculosis infection [5].

Fig. 1. Shikimate dehydrogenase-catalyzed reaction.

124 I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133

The availability of the complete genome sequence of M.tuberculosis allowed the identification of molecular targetsagainst which new therapeutical agents may be developed[6]. Homologues to the seven enzymes of the shikimatepathway were identified in the genome sequence of M.

tuberculosis. This pathway has been shown to be essentialfor the viability of M. tuberculosis [7]. Accordingly, theessentiality of the mycobacterial shikimate pathway andits absence from human host indicate that any enzyme ofthis pathway represents a promising target for the develop-ment of non-toxic antimycobacterial agents. However, anunderstanding of the kinetic mechanism and availabilityof structural models are needed to help the rational inhib-itor design to progress more rapidly [8].

Shikimate dehydrogenase (EC 1.1.1.25), the fourthenzyme in the shikimate biosynthesis pathway, catalyzesthe NADPH-dependent reduction of 3-dehydroshikimate(DHS) to shikimate (SHK) (Fig. 1). We have previouslyreported cloning and expression of M. tuberculosis SD(MtbSD) [9]. In addition, measurements of the NADP+-de-pendent oxidation of shikimate to 3-dehydroshikimate cat-alyzed by recombinant MtbSD confirmed the correctassignment to the structural gene encoding SD in M. tuber-

culosis [9]. More recently, we have reported purification tohomogeneity of recombinant functional MtbSD, N-termi-nal amino acid sequencing, electrospray ionization massspectrometry analysis, chromatographic size exclusion,and enzyme activity measurements that unambiguouslydemonstrated the identity of recombinant MtbSD, its olig-omeric state, the apparent kinetic parameters for all sub-strates, the thermal stability, and the activation energyfor the enzyme-catalyzed chemical reaction [10].

In the present work, initial velocity patterns in the for-ward direction, product inhibition studies, primary deuteri-um kinetic isotope effects, solvent kinetic isotope effects,proton inventory, multiple kinetic isotope effects, pH rateprofiles, and homology modeling were performed to inves-tigate the MtbSD kinetic and chemical mechanisms. Thedata here presented provide a framework on which to basethe rational design of inhibitors with possible antitubercu-lar activity.

Materials and methods

Enzymatic assay for MtbSD

All enzyme activity assays were carried out at 25 �C in 100 mMpotassium phosphate (pH 7.3) by monitoring the decrease in absorbance

at 340 nm (e = 6220 M�1 cm�1) accompanying the conversion of NADPHto NADP+ in the presence of DHS and MtbSD. The final enzyme con-centration was 1.8 nM. To obtain initial velocity measurements, ratevalues were determined under experimental conditions in which less than5% of substrate was consumed.

Initial velocity and product inhibition

To determine the steady-state kinetic parameters and initial velocitypatterns, MtbSD activity was measured in the presence of varying con-centrations of one substrate and several fixed-varied concentrations of theother. Product inhibition patterns were determined by measuring initialrates at five concentrations of one substrate, fixed non-saturating con-centration of the co-substrate, and fixed-varying levels of product (eitherNADP+ or SHK).

Kinetic isotope effects and proton inventory

The synthesis of [4S–2H]NADPH was performed as described byOttolina et al. (1989) [11]. Both NADPH and NADPD substrateswere purified on an FPLC Mono-Q column (GE Healthcare) aspreviously described, and the fractions with absorbance ratiosA260 nm/A340 nm 6 2.3 were pooled [12]. All measurements were per-formed in duplicate. Primary deuterium kinetic isotope effects weredetermined by measuring initial rates in the presence of varyingconcentrations of one substrate and five fixed concentrations of theother, with either NADPH or NADPD. Solvent kinetic isotope effectswere determined by measuring initial velocities using a saturatinglevel of one substrate and varying concentrations of the other ineither H2O or 90 atom% D2O. Multiple kinetic isotope effects wereobtained by determining the primary isotope effects using D2O assolvent. The proton inventory was determined using saturating con-centrations of DHS and NADPH at various mole fractions of D2O.Each proton inventory individual initial rate datum is the average oftriplicate measurements.

pH-Rate profiles

To determinate the pH dependence of kcat and kcat/Km, initial velocitieswere measured in the presence of varying concentrations of one substrateand a saturating level of the other at different pHs in a mixture ofpotassium phosphate and boric acid. The mixed buffers were used over thefollowing pH values: 5.5, 6.0, 6.5, 7.0, 7.5, 8.0, 8.5, 9.0, 9.5, and 10.0. TheMtbSD enzyme had been incubated in buffers at all the pH values men-tioned above and assayed under standard conditions to ensure enzymestability at the tested pHs.

Data analysis

Initial velocity kinetic data were fitted to appropriate equations byusing the nonlinear regression function of SigmaPlot 2000 (SPSS, Inc.).Substrate hyperbolic saturation curves at a single concentration of thefixed substrate and varying concentrations of the other substrate werefitted to Eq. (1). Intersecting initial velocity patterns were fitted to Eq. (2),which describes a sequential mechanism. For Eqs. (1) and (2), v isthe measured reaction velocity, V is the maximal velocity, A and B are theconcentrations of substrates (DHS and NADPH), Ka, and Kb are thecorresponding Michaelis–Menten constants, and Kia is the dissociationconstant for substrate A.

v ¼ VA=ðKa þ AÞ ð1Þv ¼ VAB=ðKaBþ KbAþ K iaKb þ ABÞ ð2Þ

Competitive and noncompetitive inhibition data were fitted to,respectively, Eqs. (3) and (4). For Eqs. (3) and (4), I represent the inhibitorconcentration, and Kis and Kii are the slope and intercept inhibitionconstants, respectively.

I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133 125

v ¼ VA=½Kað1þ I=K isÞ þ A� ð3Þv ¼ VA=½Kað1þ I=K isÞ þ Að1þ I=K iiÞ� ð4Þ

The kinetic isotope effect data were fitted to Eq. (5) or (6), whichdescribe, respectively, effects on both V and V/K and on V only, whereFi represents the fraction of isotopic label, and EV/K is the isotope effect onV/K minus one and EV is the isotope effect on V minus one.

v ¼ VA=½Kð1þ FiEV =KÞ þ Að1þ FiEV Þ� ð5Þv ¼ VA=½K þ Að1þ FiEV Þ� ð6Þ

Data for pH profiles data that showed a decrease in logkcat andlogkcat/Km with a slope of �1 as the pH values increased were fitted to Eq.(7), where y is the apparent kinetic parameter, C is the pH-independentplateau value of y, H is the hydrogen ion concentration, and Kb is theapparent base dissociation constant for ionizing groups.

log y ¼ log½C=ð1þ Kb=HÞ� ð7Þ

Fig. 2. Initial velocity patterns for MtbSD with both substrates DHS (A)and NADPH (B) as variable substrate. Each curve represents varied-fixedlevels of the cosubstrate. Both DHS and NADPH concentrations variedfrom 5 to 200 lM. One unit of enzyme activity (U) is defined as theamount of enzyme catalyzing the conversion of 1 lmol of NADP+ perminute at 25 �C.

Table 1Product inhibition patterns for M. tuberculosis shikimate dehydrogenasea

Variedsubstrate

Productinhibition

Inhibitiontypeb

Kis (lM)c Kii (lM)d

DHS SHK C 62.9 ± 10.5NADPH SHK NC 256.3 ± 35.4 142.5 ± 74.0DHS NADP+ NC 30.4 ± 1.3 28.3 ± 7.6NADPH NADP+ C 8.3 ± 1.3

a At 25 �C and 100 mM potassium phosphate, pH 7.3.b C = competitive, NC-MT = noncompetitive mixed-type.c Kis is the slope inhibition constant.d Kii is the intercept inhibition constant.

Homology modeling

The coordinates were from the Protein Data Bank (PDB) [13] oftemplates identified by the Blastp program [14]. Multiple sequence align-ment comparisons were carried out using ClustalW [15]. The programMODELLER6v2 [16] was used to build the protein models, followingstandard protocols for comparative protein modeling [17], and PRO-CHECK [18] used to analyze their stereochemical quality. Validation ofthe 3D profiles of the models was performed with VERIFY 3D [19]. Thestructure prediction of MtbSD here described was based on 3D structurefor the homologous SD from Escherichia coli (E. coli) (PDB ID: 1NYT),experimentally determined by X-ray diffraction at 1.5 A resolution [20].Protein structure drawing and root-mean square deviation (RMSD)between template and model structures was calculated using SwissPdb-Viewer [21].

Results

Initial velocity and product inhibition patterns

The constants for steady-state kinetics of MtbSD weredetermined under initial velocity experimental conditions.When DHS was varied at fixed-varied NADPH concentra-tions, the lines intersected on the x-axis and on the left ofthe y-axis (Fig. 2A); when NADPH was varied withfixed-varied DHS concentrations, the lines intersectedbelow the x-axis and on the left of the y-axis (Fig. 2B).The data were fitted to Eq. (2), and yielded thefollowing values: KDHS = 44 ± 3 lM, KiaDHS = 30 ± 9 lM,KNADPH = 34 ± 2 lM, kcat = 78 ± 2 s�1, kcat/KDHS =1.8 · 106 M�1 s�1, and kcat/KNADPH = 2.3 · 106 M�1 s�1.

Product inhibition experiments were performed and thedata fitted to Eq. (3) or (4). The results are given in Table 1.NADP+ is a competitive inhibitor (C) and a noncompeti-tive inhibitor (NC) versus NADPH and DHS, respectively.For SHK product, the inhibition pattern is C and NCversus, respectively, DHS and NADPH.

Kinetic isotope effects and proton inventory

The primary deuterium kinetic isotope effects on V/Kand V for both substrates are presented in Table 2. Theprimary kinetic isotope effect values of 1.8 for DVDHS

and 1.5 for DVNADPH using [4S–2H]NADPH indicate that

the C4-proS hydride (B side) is transferred to DHS in theoxy-reduction reaction catalyzed by MtbSD.

The D(V/Kapp)DHS values decreased to a limiting valueof 1.0 ± 0.03 as the concentrations of either NADPH orNADPD increased (Fig. 3), whereas D(V/Kapp)NADPH

Table 2Kinetic isotope effects for M. tuberculosis shikimate dehydrogenase

Parameter Isotope effecta

DV/KDHS 1.0 ± 0.03DVDHS 1.8 ± 0.09DV/KNADPH 1.4 ± 0.3DVNADPH 1.5 ± 0.1D2OV =KDHS 1.0 ± 0.01D2OV DHS 1.5 ± 0.3D2OV =KNADPH 1.0 ± 0.03D2OV NADPH 1.3 ± 0.2DV =KDHSðD2OÞ 1.3 ± 0.1DV DHSðD2OÞ 2.6 ± 0.01DV =KNADPHðD2OÞ 1.5 ± 0.1DV NADPHðD2OÞ 2.5 ± 0.1

a Value ± standard error obtained from fitting the data to the appro-priated equation.

126 I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133

(1.4 ± 0.3) was independent of fixed-varied concentrationsof DHS (Fig. 3 – inset).

Solvent kinetic deuterium isotope effects were deter-mined at pH 7.3 (Figs. 4A and B), a pH region in whichthe kinetic parameters are independent of small changesin pH. Small solvent kinetic isotope effects on V wereobserved for DHS (1.5 ± 0.3) and NADPH (1.3 ± 0.2),whereas no solvent effects on V/K were observed (Table2). The proton inventory, a relationship between V andthe mole fraction of D2O, was linear for MtbSD(Fig. 4A – inset).

In order to distinguish between stepwise and concertedmechanisms, multiple isotope effects were evaluated mea-suring the primary isotope effects for both substrates ineither D2O (Figs. 5A and B) or H2O (Figs. 5A and B –insets), and the results are summarized in Table 2. TheDV/KNADPH value in H2O (1.4 ± 0.3) was within standard

Fig. 3. Dependence of the apparent DV/KDHS values on the concentration of tdecay, which yielded a limiting value of 1.0 ± 0.03 for the DV/KDHS. The insetcosubstrate.

error of the DV/KNADPH value in D2O (1.5 ± 0.1), whereasDVNADPH in D2O (2.5) increased as compared to the valueobtained in H2O (1.5). The magnitude of the DVDHS valuein D2O (2.6) also increased, and, most important, runningthe reaction in D2O allowed the observation of a small, butsignificant, DV/KDHS, (1.3 ± 0.1) in contrast to the unityobtained in H2O.

pH rate profiles

To probe the role of acid/base chemistry in the mecha-nism of MtbSD, the pH dependence of kcat and kcat/Km

for DHS and NADPH was determined over the pH rangeof 5.5 6 pH 6 10.0. The pH profiles of kcat for both sub-strates are very similar and decrease at high pHs with aslope of �1 (Fig. 6A), demonstrating that deprotonationof a single residue with apparent pKa value of 8.9 ± 0.1abolishes MtbSD catalytic activity. The kcat/Km profilesfor DHS and NADPH show a decrease at high pH valueswith a slope of �1 (Figs. 6B and C), suggesting that depro-tonation of a single ionizable group with pKa value of9.1 ± 0.1 diminishes DHS and NADPH binding.

Homology modeling

E. coli SD (PDB ID: 1NYT) [20], E. coli YdiB (PDB ID:109B) [20] and Methanococcus jannaschii SD (PDB ID:1NVT) [22] were identified as candidate structural tem-plates for homology modeling. Although Blastp results list-ed M. jannaschii SD as the best template, the structurecoordinates were not available. The second best result,E. coli YdiB, is a dual specificity quinate/shikimate dehy-drogenase that utilizes either NAD+ or NADP+, required

he cosubstrate. The data were fitted to an equation describing a hyperbolicshows that DV/KNADPH values do not depend on the concentration of the

Fig. 4. Solvent isotope effects for MtbSD. (A) DHS as varied substrate,with saturating concentration of the cosubstrate. Reaction mix containedeither 0 (d) or 90 (.) atom% D2O. Inset represents the proton inventoryon MtbSD, with both substrates at saturating concentrations. (B)NADPH as varied substrate, with saturating concentration of thecosubstrate. Reaction mix contained either 0 (d) or 90 (.) atom% D2O.

Fig. 5. Multiple isotope effects for MtbSD. (A) NADPD as variedsubstrate, with saturating concentration of DHS. Reaction mix containedeither 0 (d) or 90 (.) atom% D2O. Inset represents the primary isotopeeffect, NADPH (d) or NADPD (.) as varied substrate with saturatingconcentration of the DHS. (B) DHS as varied substrate, with saturatingconcentration of NADPD. Reaction mix contained either 0 (d) or 90 (.)atom% D2O. Inset represents the primary isotope effect, DHS as variedsubstrate with saturating concentration of NADPH (d) or NADPD (.).

I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133 127

insertion of two gap regions into the MtbSD sequence(Fig. 7). However, segments of the target sequence whichhave no equivalent in the template are the most difficultregions to model. Hence, the E. coli SD (PDB ID:1NYT), solved experimentally by X-ray diffraction at1.5 A resolution [20], was used as template to model the3D structure of MtbSD, even though they have approxi-mately 25% sequence identity, the limit usually allowedfor comparative protein structure modeling [14]. Ten mod-els of the enzyme were built and evaluated to choose thebest one. MtbSD contains 269 amino acid residues. Outof 220 non-glycine and non-proline residues, 199% or90.5% were located in the most favored regions of theRamachandran plot. The best model of MtbSD (Fig. 8)has a backbone RMSD value of 1.93 A. Considering thesequence divergence between these ortholog sequences,approximately 25% sequence identity, this RMSD valueis expected and the modeling results, overall, demonstratethat the MtbSD model obtained is highly satisfactory andcan thus be used to infer structure-activity relationships.

Discussion

Kinetic mechanism of MtbSD

The families of intersecting double-reciprocal plotsobserved in the initial velocity for the forward directionfor both substrates are consistent with a sequential mecha-nism and thus both substrates must attach to the enzymeforming a ternary complex before any product is released.Accordingly, a ping-pong mechanism could be discarded.In addition, a rapid equilibrium ordered mechanism wasruled out because the families of double reciprocal plotsobtained intersect on the left of the y-axis (Figs. 2A and B).

The steady-state kinetic constant values given here arelarger than the apparent kinetic parameters we have previ-ously reported [10], whose data were collected in Tris–HCl100 mM, pH 7.0. We verified a significant reduction in thecatalytic rate for this enzyme when the assay mix was buf-fered using Tris–HCl instead of potassium phosphate (data

Fig. 6. Dependence of MtbSD kinetics parameters on pH. (A) pHdependence of logkcat, (B) pH dependence of logkcat/KDHS, and (C) pHdependence of logkcat/KNADPH. Experimental data were fitted to Eq. (7).

128 I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133

not shown). In addition, here we report true steady-statekinetic parameters instead of apparent kinetic parameters.

Determination of steady-state kinetic parameters forPisum sativum SD [23] in 100 mM potassium phosphate,pH 7.4, yielded values of 340 ± 40 lM for KDHS, which isapproximately 8-fold larger than the value for MtbSD(44 lM), and 4.3 ± 0.5 lM for KNADPH, which is approxi-mately 8-fold smaller than the value here reported forMtbSD (34 lM).

The first step in determining the mechanism of action ofan enzyme is to establish its kinetic mechanism, the order

of substrate binding and release of products. This is mostcommonly achieved by product inhibition studies in whichwe observe the patterns obtained when substrate concen-trations are varied at fixed levels of other substrate(s) inthe presence of products, yielding information not avail-able from initial velocity studies.

Product inhibition data for MtbSD (Table 1) were sim-ilar to those for P. sativum SD. For the reaction of P. sat-

ivum SD [23], NADPH was found to give linearcompetitive inhibition with NADP+ as variable substrate,and NADPH was a noncompetitive inhibitor versus SHKat low NADP+ concentrations. The results for SHK inhibi-tion versus DHS indicated slightly noncompetitive inhibi-tion, thereby suggesting that the mechanism is ordered[23]. Product inhibition data presented here suggest twopossible kinetic mechanisms for MtbSD, steady-stateordered and rapid equilibrium random with two dead-end ternary complexes (MtbSD-NADPH–SHK andMtbSD-NADP+–DHS). An application of primary deute-rium isotope effects is to distinguish among possible kineticmechanisms. Accordingly, to differentiate between steady-state ordered and rapid equilibrium random, analysis ofprimary kinetic isotope effects was carried out.

Measurements of the steady-state primary deuteriumkinetic isotope effects on apparent V/K for one substratewere carried out at various cosubstrate concentrations(Fig. 3). Based on the mechanistic deductions from isotopeeffects for multireactant enzymes developed by Cook andCleland [24], the primary isotope effect on D(V/Kapp)B isindependent of the concentration of A and D(V/Kapp)A

decreases as the concentration of B increases and reaches alimiting value of 1.0 at infinite concentration of B for asteady-state ordered mechanism. A and B refer to, respec-tively, the first and second substrates to bind to the enzyme.The kinetic isotope effects on V/K results (Fig. 3) are in agree-ment with a steady-state ordered bi-bi mechanism with DHSbinding first followed by NADPH binding to MtbSD activesite (Fig. 9). In addition, the DHS substrate is sticky, hence itreacts to give SHK as fast as, or faster than, it dissociatesfrom the enzyme [24]. Based on initial velocity patterns,product inhibition, analysis of Haldane relationships andisotope-exchange studies, an ordered kinetic mechanismwith NADPH binding first has been proposed for P. sativum

SD [23]. Initial velocity and dead-end inhibition studiesshowed that the kinetic mechanism is ordered bi-bi formouse class II alcohol dehydrogenase with coenzyme bind-ing first [25]. The NAD-dependent mouse class II alcoholdehydrogenase is structurally homologous to E. coli SDdinucleotide-binding domain [20].

Stereospecificity and rate-limiting steps

The primary deuterium kinetic isotope effects indicatethat the C4-proS hydride (B side) is transferred to DHSin the oxy-reduction reaction catalyzed by MtbSD. In con-trast, E. coli SD has been shown to transfer the C4-proR

hydride (A-side) of NADPH [26], in agreement with the

Fig. 7. ClustalW multiple sequence alignment between the target (MtbSD) and templates (E. coli YdiB and E. coli SD). YdiB is more similar to MtbSDthan to E. coli SD and optimal alignment between the three sequences requires the insertion of two gap regions in the C-terminus of MtbSD and E. coli SDsequences (underlined in YdiB). a-helices and b-strands are represented as H and S, respectively. This sequence alignment was created using the followingsequences from GeneBank�: M. tuberculosis SD (CAB06186, residues 1–269), E. coli K12 YdiB (NP 416207, residues 1–288), and E. coli K12 SD(NP_417740, residues 1–272).

Fig. 8. Stereo ribbon representation of the MtbSD structure. The MtbSD structure is characterized by a nearly canonical Rossmann fold with a six-stranded parallel b sheet (C-terminal domain on top), and a a/b N-terminal domain (bottom) with an antiparallel b-strand (b5). The a-helices are shown inlight gray and the b-strands in black.

I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133 129

Fig. 9. Proposed kinetic mechanism for M. tuberculosis shikimate dehydrogenase.

130 I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133

crystal structure of the enzyme [20]. Isotope effects on V

report on events following formation of the ternary com-plex capable of undergoing catalysis, which include thechemical steps, possible enzyme conformational changes,and product release [27]. Isotope effects on V/K report onsteps in the reaction mechanism from the binding of theisotopically labeled substrate to the first irreversible step,usually the release of the first product. The apparent clas-sical limit for primary deuterium kinetic isotope effects onthe maximal velocity is around 8 if carbon-hydrogen bondcleavage is the rate-determining step, even though values assmall as 2 have been used, in a less rigorous practice, as evi-dence for rate-limiting steps [27]. The primary kinetic iso-tope effect values of 1.8 for DVDHS and 1.5 for DVNADPH

using [4S–2H]NADPH as reductant indicate that thehydride transfer is only partly rate limiting for MtbSDenzyme catalysis. Isotope-insensitive steps such as enzymeisomerization and/or product release are probably contrib-uting to the rate-limiting steps of the MtbSD reaction. TheD(V/Kapp)NADPH and D(V/Kapp)DHS values indicate thatsubstrate binding steps make a small contribution to therate-limiting steps.

Steady-state solvent isotope effects were evaluated toassess the contribution of solvent proton transfer to a stepin the enzymatic mechanism. The effect on V arises fromsolvent-exchangeable protons being transferred duringcatalysis. The values for solvent isotope effect on V forDHS (1.5) and NADPH (1.3) indicate that solvent-ex-changeable protons being transferred during catalysis isonly partly rate limiting in the overall reaction. No solventisotope effect on V/K was observed for DHS and NADPH.

The number of protons transferred during the oxy-re-duction reaction catalyzed by MtbSD was determined bythe proton inventory technique. Measurements of V in dif-ferent isotopic solvent mixtures (V relative versus mol frac-tion of D2O) showed a linear relationship (Fig. 4A – inset),suggesting that a single proton is transferred in the stepthat exhibits the solvent isotope effect [28]. A similar resultwas observed for the M. tuberculosis NADPH-dependentmycothione reductase [29].

Chemical mechanism of MtbSD

Double isotope effect studies are able to distinguishwhether two different isotopic substitutions affect the sameor different chemical steps. Assuming that the primary deu-terium kinetic isotope effect is expressed only on thehydride transfer reaction (Figs. 5A and B – insets) and sol-vent isotope effects affect only the alkoxide protonation,double isotope effects can distinguish whether theseisotopic substitutions affect the same or different chemicalsteps (Figs. 5A and B). Theory predicts that if protonationand hydride transfer occur in the same transition state, theprimary isotope effects will be larger or unchanged with

D2O as compared to H2O. On the other hand, if hydridetransfer and protonation occur in distinct steps, theprimary isotope effects will be smaller with D2O as sol-vent, as proton transfer will become more rate limiting[30,31]. The increased values for the primary isotope effectson V/K measured in D2 OðDV =KDHSðD2OÞ ¼ 1:3 andDV =KNADPHðD2OÞ ¼ 1:5Þ as compared to values measuredin H2O (DV/KDHS = 1.0 and DV/KNADPH = 1.4) are inagreement with both hydride and proton transfer takingplace in the same step (concerted mechanism), thus imply-ing a single transition state. It should be pointed out thatuse of D2O allowed the observation of a significant valuefor DV/KDHS (1.3), even though this substrate is stickyand reaction proceeds through a steady-state orderedkinetic mechanism.

The b-elimination catalyzed by bovine liver crotonase isconcerted [32], as well as are the reactions catalyzed by pigliver acyl-CoA dehydrogenase [33] and isocitrate dehydro-genase [34]. Concerted reactions are a common strategyutilized by enzymes to avoid unstable intermediates, suchas some enolate anions [35,36].

The role of acid/base chemistry in the mechanism ofMtbSD was determined using the pH dependence of kcat

and kcat/Km for DHS and NADPH in pH range of5.5 6 pH 6 10.0 (Fig. 6). In this experiment the kcat forboth substrates are very similar and decrease at high pHswith a slope of �1 (Fig. 6A), indicating that deprotonationof a single residue with apparent pKa value of 8.9 ± 0.1abolishes MtbSD catalytic activity. The kcat/Km profilesfor DHS and NADPH show that deprotonation of a singleionizable group with pKa value of 9.1 ± 0.1 abolishes theability of DHS and NADPH to bind and react (Figs. 6Band C). It is likely that the ionization behavior of the samegroup is being observed in the pH profiles for kcat and kcat/Km, and the difference in the pK values may reflect pertur-bation of the pK value upon substrate(s) binding toMtbSD. The pK values for kcat and kcat/Km lie in the nor-mal pK range for e-amino of lysine, thiol of cysteines andphenolic hydroxyl of tyrosine amino acids. The pH-rateprofiles of Haemophilus influenzae SD for the reverse reac-tion indicated that a group with pK value of �8.1 needs tobe deprotonated for activity, and site-directed mutagenesisresults were consistent with Lys-67 playing a key role incatalysis [37].

Based on the double isotope effects and pH-rate profiles,we propose a chemical mechanism for MtbSD (Fig. 10), inwhich hydride transfer and solvent proton transfer are con-certed, and an amino acid residue with pKa value of 8.9 isinvolved in catalysis.

Analysis of the structure of E. coli SD shows that DHSis positioned for stereospecific reduction to SHK and theside chain of Lys65 has been proposed to hydrogen bondto the C-4 hydroxyl group of DHS/SHK (E.coli SD num-bering) [20]. This amino acid residue has been proposed

Fig. 10. Proposed chemical mechanism for M. tuberculosis shikimate dehydrogenase-catalyzed reaction. R represents ribose, adenosine diphosphate, and2 0-phosphate moieties of NADPH.

I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133 131

to be the acid/base catalytic group that donates a proton tothe carbonyl of DHS during reduction and that removes aproton during oxidation of SHK [20]. In agreement, previ-ous studies of P. sativum SD showed that substrate-likeinhibitors of the enzyme require a C-4 hydroxyl, whereaseither a C-5 hydroxyl or carboxylate group is needed forstrong binding [38]. Analogs of DHS substrate that lackthe C-4 and C-5 hydroxyls were used to demonstrate therole of both C-5 hydroxyl and C-4 hydroxyl on the sub-strate specificity for the E. coli SD, and the results showedthat the C-4 hydroxyl has a very significant effect on thespecificity of the substrate [39]. It has been suggested thatthe C-4 hydroxyl group hydrogen bonds to a chargedgroup in the E. coli SD active site based on an estimationof the binding energy given by kcat/Km [39]. However, itshould be kept in mind that participation of a lysine sidechain residue in the chemical reaction catalyzed by E. coli

SD is based on a molecular model, since the crystal struc-ture is for E. coli SD-NADP+ binary complex, and E. coli

SD transfers the C4-proR hydride (A-side) of NADPH.Based on pH rate profiles of P. sativum SD, it has been pos-tulated that a group with pKa value of 9.4, possibly or an e-amino group of lysine, binds the carboxylate ion of thesubstrate, while a group of pKa 8.6, possibly a sulphydrylresidue, interacts with the C-4 hydroxyl group of DHS/SHK [40]. The crystal structure of M. jannaschii SD hasbeen determined and the residues involved in DHS bindingand its catalytic reduction were identified, amongst themLys70 (M. jannaschii SD numbering) [22]. The side chainof Asp102 of E. coli SD has also been suggested to forma hydrogen bond to C-4 hydroxyl group of the substrate[20]. In addition, analysis of the crystal structure of H.

influenzae SD and site-directed mutagenesis studies haveshown that Asp103 and Lys67 (H. influenzae SD number-ing) may function as a catalytic pair involved in acid/basecatalysis for this enzyme [37]. However, the pH-rate pro-files for MtbSD did not show any residue whose proton-ation would abolish enzyme activity at low pH values.However, it could be argued that the lowest pH value(5.0) of the pH-rate profiles presented here would not allow

detection of carboxylic groups of aspartate side chainssince its pK value is approximately 4. Notwithstanding,the pK for the carboxylate group of DHQ/SHK is approx-imately 4.1 and would be a difficult task to discriminatebetween substrate and/or amino acid ionizing groups. Wehave previously shown by analysis of multiple sequencealignment of MtbSD, E. coli SD, H. influenzae SD-like,and M. jannaschii SD that Lys69 and Asp105 (M. tubercu-

losis numbering) are conserved [10]. The pH-rate profilesdescribed here show participation of a group whose depro-tonation abolishes binding (pKa = 9.1) and catalytic activ-ity (pKa = 8.9), consistent with participation of Lys69. Atany rate, site directed mutagenesis of these residues andmeasurements of the steady-state kinetic parameters forMtbSD enzyme reaction are currently underway to evalu-ate the role, if any, of Lys69 and Asp105 may play in bind-ing and/or catalysis.

MtbSD structure

The MtbSD structure belongs to the homologous super-family denominated NAD(P)-binding Rossmann-folddomain [41]. The SDH family provides a new example ofa protein family displaying the dinucleotide binding-fold,without significant sequence homology with other Ross-mann-fold families [20]. The structure has an elongatedshape with two domains (Fig. 8). Each domain has an a/b architecture. The domains are linked by an a-helix anda turn that keeps them together and form a deep grooveinto which the NADP cofactor binds. The NADP-bindingdomain, in the C-terminus, adopts a nearly canonical Ross-mann fold, a six-stranded parallel b-sheet with six a-heli-ces, three on each side of the b-sheet. The N-terminusconsists of a six-stranded b-sheet and six a-helices, howeverthe arrangement is irregular, since the near central b5strand is in an antiparallel orientation with respect to theother strands [20].

The substrate-binding site of E. coli SD, which was iden-tified by the position of the nicotinamide ring, is delineatedby the SDH-family conserved residues Ser14, Ser16, Lys65,

132 I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133

Asn86, Thr101, Asp102, and Gln244 [20]. These residueswere identified in the MtbSD model (Ser18, Ser20, Lys69,Asn90, Thr104, Asp105, and Gln243; MtbSD numbering)(Fig. 11). The E. coli SD conserved Ser14, Ser16 andTyr215 have been suggested to be involved in substrate car-boxylate binding [20]. These residues were found in theMtbSD model (Fig. 11) to be in equivalent positions andappear to play a role in binding of the carboxylate groupof DHS/SKH.

In the multiple sequence alignment between the target(MtbSD) and templates (E. coli YdiB and E. coli SD) weidentified the sequence Gly124-Ser125-Gly126-Gly127-Thr128-Ala129 as MtbSD sequence pattern (G [A,s,g] GG [A,t] [A,S,g]), corresponding to the diphosphate-bindingloop, which is conserved in the entire SDH family [20](Fig. 7). Interaction between the diphosphate-binding loopand the active site in the E. coli SD occurs through Gly129and Ala130 residues, which correspond to, respectively,Thr128 and Ala129 in MtbSD (Fig. 7). For the bindingof the nicotinamide in E. coli SD it was verified that theamide group N-7 of the nicotinamide ring is hydrogen-bonded to the carbonyl group of Met213 and of theinvariant Gly237 [20]. In the MtbSD model, the Gly236(corresponding to E.coli SD Gly237) is hydrogen-bondedto the amide group N-7. On the other hand, the MtbSDmodel presents Ala213 residue in equivalent position of

Fig. 11. The conserved residues in the nicotinamine ring pocket in the structuserine residues involved in hydrogen-bond with the DHS carboxylate group areand in the MtbSD (Tyr215, Ser18, and Ser20; gray).

the E. coli Met213. Notwithstanding, an equivalent bondwith the amide group N-7 of nicotinamide ring is main-tained with MtbSD Ala213. The E. coli SD Arg150 andArg154 play a crucial role in adenine phosphate bindingas they form an ‘‘electrostatic clamp’’ that sandwiches thephosphate substituent of NADP+ [20]. In the MtbSD modelit was found that Arg149 and Lys153 residues occupy equiv-alent positions, and it is thus tempting to suggest that theseresidues play a role in adenine phosphate binding in MtbSD.

The work here presented is, to the best of our knowl-edge, the first detailed report on the steady-state velocitypatterns, product inhibition, primary deuterium kinetic iso-tope effects, solvent kinetic isotope effects, proton invento-ry, double isotope effects, pH-rate profiles, and molecularmodeling of MtbSD. Initial velocity patterns in the forwardreaction, product inhibition studies, and primary deuteri-um kinetic isotope effects allowed us to propose a steady-state ordered bi-bi kinetic mechanism for catalysis, withDHS binding first followed by NADPH binding to theMtbSD enzyme catalytic site. The primary deuteriumkinetic isotope effects indicated that the C4-proS hydrideis transferred from the NADPH co-substrate in a step thatis only partly rate limiting. Solvent kinetic isotope effectsdemonstrated that proton transfer from the solvent is onlypartly rate limiting. Proton inventory results indicated thata single proton is transferred in the solvent-sensitive step.

res of both E. coli SD (black) and in the MtbSD (gray). The tyrosine andin equivalent positions in both E. coli SD (Tyr215, Ser14, and Ser16; black)

I.O. Fonseca et al. / Archives of Biochemistry and Biophysics 457 (2007) 123–133 133

Double isotope effects showed that transfer of hydride andproton occur in concert. The pH-rate profiles revealed thata charged group, probably the e-amino group of Lys69plays an important role in catalysis and substrate binding.The homology 3D model for MtbSD identified theprobable residues in the substrate binding site, amongstthem Lys69, in agreement with pH studies, and Asp105.These results allowed us to propose a kinetic and chemicalmechanism for MtbSD. The enzyme kinetics and modelingstudies provide a framework on which to base the design ofenzyme inhibitors with potential use as antitubercularagents. Site-directed mutagenesis, equilibrium bindingspectrofluorimetry, and pre-steady state experiments arecurrently underway to improve our understanding of themechanism of action of MtbSD.

Acknowledgments

Financial support for this work was provided byMillennium Initiative Program MCT-CNPq, Ministry ofHealth-Department of Science and Technology (DECIT)-UNESCO (Brazil) to D.S.S. and L.A.B. D.S.S. andL.A.B. also acknowledge grants awarded by CNPq,FINEP, and PRONEX/FAPERGS/CNPq. D.S.S. (CNPq,304051/1975-06) and L.A.B. (CNPq, 520182/99-5) areresearch career awardees from the National ResearchCouncil of Brazil. We thank Professor John W. Frost,Department of Chemistry of Michigan State University,for his generous gift of 3-dehydroshikimate substrate.

References

[1] C. Dye, S. Scheele, P. Dolin, V. Pathania, M.C. Raviglione, JAMA282 (1999) 677–686.

[2] L.A. Basso, J.S. Blanchard, Adv. Exp. Med. Biol. 456 (1998) 115–144.[3] A. Plabos-Mendez, D.K. Gowda, T.R. Frieden, World Health Organ.

80 (2002) 489–495.[4] CDC (Centers for Disease Control and Prevention), Morb. Mortal.

Wkly. Rep. 55 (2006) 301–305.[5] R.J. O’Brien, P.P. Nunn, Am. J. Respir. Crit. Care Med. 162 (2001)

1055–1958.[6] S.T. Cole, R. Brosch, J. Parkhill, T. Garnier, C. Churcher, D. Harris,

S.V. Gordon, K. Eiglmeier, S. Gas, C.E. Barry 3rd, F. Tekaia, K.Badcock, D. Basham, D. Brown, T. Chillingworth, R. Connor, R.Davies, K. Devlin, T. Feltwell, S. Gentles, N. Hamlin, S. Holroyd, T.Hornsby, K. Jagels, A. Krogh, J. McLean, S. Moule, L. Murphy, K.Oliver, J. Osborne, M.A. Quail, M.A. Rajandream, J. Rogers, S.Rutter, K. Seeger, J. Skelton, R. Squares, S. Squares, J.E. Sulston, K.Taylor, S. Whitehead, B.G. Barrell, Nature 393 (1998) 537–544.

[7] T. Parish, N.G. Stoker, Microbiology 148 (2002) 3069–3077.

[8] J.R. Coggins, C. Abell, L.B. Evans, M. Frederickson, D.A. Robin-son, A.W. Roszak, A.P. Lapthorn, Biochem. Soc. Trans. 31 (2003)548–552.

[9] M.L. Magalhaes, C.P. Pereira, L.A. Basso, D.S. Santos, ProteinExpr. Purif. 26 (2002) 59–64.

[10] I.O. Fonseca, M.L.B. Magalhaes, J.S. Oliveira, R.G. Silva, M.A.Mendes, M.S. Palma, D.S. Santos, L.A. Basso, Protein Expr. Purif.46 (2006) 429–437.

[11] G. Ottolina, S. Riva, G. Carrea, B. Danieli, A.F. Buckmann,Biochem. Biophys. Acta 998 (1989) 173–178.

[12] G.A. Orr, J.S. Blanchard, Anal. Biochem. 142 (1984) 232–234.[13] H.M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T.N. Bhat, H.

Weissig, I.N. Shindyalov, P.E. Bourne, Nucleic Acids Res. 28 (2000)235–242.

[14] S.F. Altschul, T.L. Madden, A.A. Schaffer, J. Zhang, Z. Zhang, W.Miller, D.J. Lipman, Nucleic Acids Res. 25 (1997) 3389–3402.

[15] J.D. Thompson, D.G. Higgins, T.J. Gibson, Nucleic Acids Res. 22(1994) 4673–4680.

[16] A Sali, T.L. Blundell, J. Mol. Biol. 234 (1993) 779–815.[17] M.A. Martı-Renom, A.C. Stuart, A. Fiser, R. Sanchez, F. Melo, A.

Sali, Annu. Rev. Biophys. Biomol. Struct. 29 (2000) 291–325.[18] R.A. Laskowski, M.W. MacArthur, D.S. Moss, J.M. Thornton, J.

Appl. Cryst. 26 (1993) 283–291.[19] R. Luthy, J.U. Bowie, D. Eisenberg, Nature 356 (1992) 83–85.[20] G. Michel, A.W. Roszak, V. Sauve, J. Maclean, A. Matte, J.R.

Coggins, M. Cygler, A.J. Lapthorn, J. Biol. Chem. 278 (2003) 19463–19472.

[21] N. Guex, M.C. Peitsch, Electrophoresis 18 (1997) 2714–2723.[22] A.K. Padyana, S.K. Burley, Structure 11 (2003) 1005–1013.[23] D. Balinsky, A.W. Dennis, W.W. Cleland, Biochemistry 10 (1971)

1947–1952.[24] P.F. Cook, W.W. Cleland, Biochemistry 20 (1981) 1790–1796.[25] P. Stromberg, S. Svensson, K.B. Berst, B.V. Plapp, J.O. Hoog,

Biochemistry 43 (2004) 1323–1328.[26] P. Dansette, R. Azerad, Biochimie 56 (1974) 751–755.[27] D.B. Northrop, Biochemistry 14 (1975) 2644–2651.[28] D.M. Quinn, L.D. Sutton, in: P.F. Cook (Ed.), Enzyme Mechanism

from Solvent Isotope Effects, CRC Press, Florida, 1991, pp. 73–126.[29] M.P. Patel, J.S. Blanchard, Biochemistry 40 (2001) 5119–5126.[30] J.D. Hermes, C.A. Roeske, M.H. O’Leary, W.W. Cleland, Biochem-

istry 21 (1984) 5106–5114.[31] J.G. Belasco, J. Albery, J.R. Knowles, J. Am. Chem. Soc. 105 (1983)

2475–2477.[32] J.A. Gerlt, Bioorganic Chemistry: Peptides and Proteins, Oxford

University Press, New York, 1998.[33] B.J. Bahnson, V.E. Anderson, Biochemistry 30 (1991) 5894–5906.[34] B. Pohl, T. Raichle, S. Ghisla, Eur. J. Biochem. 160 (1986) 109–115.[35] P.F. Cook, W.W. Cleland, Biochemistry 20 (1981) 1797–1805.[36] A. Thibblin, W.P. Jencks, J. Am. Chem. Soc. 101 (1979) 4963–4973.[37] S. Singh, S. Korolev, O. Koroleva, T. Zarembinski, F. Collart, A.

Joachimiak, D. Christendat, J. Biol. Chem. 280 (2005) 17101–17108.[38] D. Balinsky, D.D. Davies, Biochem. J. 80 (1961) 296–300.[39] T.D. Bugg, C. Abell, J.R. Coggins, Tetrahedron Lett. 29 (1988) 6779–

6782.[40] A.W. Dennis, D. Balinsky, Int. J. Biochem. 3 (1972) 93–102.[41] A.G. Murzin, S.E. Brenner, T. Hubbard, C. Chothia, J. Mol. Biol.

247 (1995) 536–540.