JOURNAL OF BACTERIOLOGY, Sept. 2006, p. 6081–6091 Vol. 188, No. 170021-9193/06/$08.00�0 doi:10.1128/JB.00338-06Copyright © 2006, American Society for Microbiology. All Rights Reserved.

The Structure of MbtI from Mycobacterium tuberculosis, the First Enzymein the Biosynthesis of the Siderophore Mycobactin,

Reveals It To Be a Salicylate SynthaseAnthony J. Harrison,1,2 Minmin Yu,4 Theres Gårdenborg,1,2† Martin Middleditch,2

Rochelle J. Ramsay,1,2 Edward N. Baker,1,2,3 and J. Shaun Lott1,2*Centre for Molecular Biodiscovery,1 School of Biological Sciences,2 and Department of Chemistry,3 University of Auckland, Auckland,

New Zealand, and Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 947204

Received 9 March 2006/Accepted 21 June 2006

The ability to acquire iron from the extracellular environment is a key determinant of pathogenicity inmycobacteria. Mycobacterium tuberculosis acquires iron exclusively via the siderophore mycobactin T, thebiosynthesis of which depends on the production of salicylate from chorismate. Salicylate production in otherbacteria is either a two-step process involving an isochorismate synthase (chorismate isomerase) and apyruvate lyase, as observed for Pseudomonas aeruginosa, or a single-step conversion catalyzed by a salicylatesynthase, as with Yersinia enterocolitica. Here we present the structure of the enzyme MbtI (Rv2386c) from M.tuberculosis, solved by multiwavelength anomalous diffraction at a resolution of 1.8 Å, and biochemical evidencethat it is the salicylate synthase necessary for mycobactin biosynthesis. The enzyme is critically dependent onMg2� for activity and produces salicylate via an isochorismate intermediate. MbtI is structurally similar tosalicylate synthase (Irp9) from Y. enterocolitica and the large subunit of anthranilate synthase (TrpE) andshares the overall architecture of other chorismate-utilizing enzymes, such as the related aminodeoxychoris-mate synthase PabB. Like Irp9, but unlike TrpE or PabB, MbtI is neither regulated by nor structurallystabilized by bound tryptophan. The structure of MbtI is the starting point for the design of inhibitors ofsiderophore biosynthesis, which may make useful lead compounds for the production of new antituberculosisdrugs, given the strong dependence of pathogenesis on iron acquisition in M. tuberculosis.

Mycobacterium tuberculosis, the cause of tuberculosis, is adevastating human pathogen, responsible for more than twomillion deaths annually (8). Although drugs that target activelygrowing M. tuberculosis are available, their effectiveness iscompromised by the requirement for long treatment times (43)and the growing problem of multidrug resistance (15). A fur-ther problem is posed by the ability of the organism to enter anonreplicating, persistent state after engulfment by activatedmacrophages in the lung (22). It is estimated that one-third ofthe world’s population harbors a latent infection of this kindand is at risk of reactivation of disease (38).

Iron is essential for mycobacterial growth, as it is for virtuallyall living systems. For pathogenic bacteria, such as M. tubercu-losis, iron acquisition is strongly correlated with virulence (45,46). It has widely been assumed that macrophage-engulfedmycobacteria survive in an iron-poor environment within thephagosome, as has been demonstrated for Salmonella entericaserovar Typhimurium in epithelial cell vacuoles (19). However,recent studies have demonstrated that pathogenic mycobacte-ria (M. tuberculosis and Mycobacterium avium) actively increasethe iron concentration in their phagosomes (53), presumablyby capturing iron from endosomes containing transferrin (5),

an ability not shared by nonpathogenic species, such as Myco-bacterium smegmatis.

The physicochemical properties of free iron and its toxicityare such that sophisticated mechanisms have evolved to cap-ture, solubilize, and assimilate this essential element. Myco-bacteria, like most other microorganisms, synthesize chelatingmolecules called siderophores for this purpose (see references10, 49, and 55 for recent reviews). Siderophores are producedin response to iron deficiency and are secreted into the envi-ronment, where they bind iron with high affinity and transfer itback into the cell (45). The siderophores produced by myco-bacteria are of several types, notably, the salicylate-based my-cobactins and the peptidohydroxamate-based exochelins. M.tuberculosis produces only mycobactin T (Fig. 1), although it isproduced in several forms that differ in the nature of the acylside chains attached to the central modified lysine residue.Both a water-soluble form, which is secreted into the externalmedium, and an insoluble, membrane-associated form are re-quired in vivo for mycobacterial pathogenesis (13, 20, 45).

The biosynthetic pathway by which mycobactin is producedhas largely been elucidated (13, 45), although not all steps orenzymes have been specifically identified. Most of the enzymesthat are implicated in the pathway are encoded by open read-ing frames that are clustered in the putative mbt operon, ex-tending from Rv2377c to Rv2386c in the M. tuberculosisH37Rv genome (6). Deletion of mbtB (Rv2383c) showed thatthis gene, and by implication others involved in mycobactinbiosynthesis, is essential for growth in macrophages and isrequired for virulence (12). This gene knockout also destroys

* Corresponding author. Mailing address: School of Biological Sci-ences, University of Auckland, Private Bag 92-019, Auckland 1020,New Zealand. Phone: 64-9-373-7599. Fax: 64-9-373-7414. E-mail: s.lott@auckland.ac.nz.

† Present address: IBG, Uppsala Biomedicinska Center, UppsalaUniversity, Box 592, S-75124 Uppsala, Sweden.

6081

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

the ability of the bacterium to accumulate iron when engulfedin the phagosome (53).

The last gene in the mbt cluster, Rv2386c, has also beenshown, by genome-wide transposon mutagenesis (48), to beessential for the in vitro growth of M. tuberculosis. It wasoriginally annotated as trpE2 (6) because of its similarity totrpE, the gene encoding the large subunit of anthranilate syn-thase, the enzyme that catalyzes the first committed step intryptophan biosynthesis. However, like expression of othermembers of the mbt operon, the expression of Rv2386c isregulated by the iron response repressor IdeR. Under condi-tions of low iron, repression by IdeR is lost, and expression ofRv2386c and other mbt genes is induced (21, 47). The Rv2386cgene was thus reannotated as mbtI, putatively encoding theenzyme isochorismate synthase (44), which catalyzes the firststep in the formation of salicylate and ultimately mycobactin.Sequence analysis of a number of chorismate-utilizing enzymesshows that Rv2386c (mbtI) clusters more closely with pchA, thegene encoding isochorismate synthase in Pseudomonas aerugi-nosa, than it does with trpE from a number of bacterial species(18). PchA has also been shown biochemically to act as anisochorismate synthase in the first step in the biosynthesis ofthe salicylate-containing siderophore pyochelin in P. aerugi-nosa (18).

Chorismate is a key intermediate in the biosynthesis of manyessential aromatic compounds, being converted to prephenatein phenylalanine and tyrosine biosynthesis, anthranilate intryptophan biosynthesis, p-aminobenzoic acid (PABA) in fo-late biosynthesis, p-hydroxybenzoate in ubiquinone and mena-quinone biosynthesis, and salicylate in siderophore biosynthe-sis (54). The enzymes that catalyze these conversions ofchorismate share a degree of sequence identity (typically onthe order of 20% on a pairwise basis) that implies relatedstructures. Three-dimensional structures are currently avail-able for salicylate synthase (Irp9) from Yersinia enterocolitica(29), for anthranilate synthase (TrpE) from Sulfolobus solfa-taricus, Serratia marcescens, and Salmonella enterica serovarTyphimurium (31, 36, 50), and for aminodeoxychorismate syn-thase (PabB) from Escherichia coli (39). Anthranilate synthaseis a hetero-oligomeric complex composed of the products ofthe trpE and trpG genes. TrpG is a glutamine amidotransferase

which provides the amino group required in the biosynthesis ofanthranilate. Aminodeoxychorismate synthase forms an anal-ogous heterodimer, with PabA functioning as an amidotrans-ferase. In contrast, the salicylate synthase from Y. enterocoliticais homodimeric. Although the structures of TrpE, PabB, andIrp9 share a common fold (29, 39), the fine structural differ-ences that enable the production of different products by re-lated enzymes (Fig. 2) are of great interest, particularly giventhe attractiveness of these enzymes for structure-based drugdesign.

Salicylate is a key compound required for the biosynthesis ofsiderophores in a number of bacterial species, and its produc-tion from chorismate is analogous to that of anthranilate intryptophan biosynthesis or of PABA in folate biosynthesis (Fig.2). In P. aeruginosa, a two-step conversion of chorismate tosalicylate has been demonstrated, with the PchA protein actingas an isochorismate synthase (i.e., a chorismate isomerase,responsible for moving the hydroxyl group from position 4 toposition 2 on the chorismate ring) and the PchB protein actingas an isochorismate-pyruvate lyase, cleaving the pyruvate moi-ety from the ring (17, 18). This is equivalent to the situation inPABA production in many bacteria, where the PabB proteinfacilitates chemical modification of the chorismate ring (in thiscase the addition of an amino group is provided by the amido-transferase PabA) but a separate protein, PabC, provides lyaseactivity (24, 37). In contrast, in tryptophan biosynthesis, theTrpE protein is able to act as both ring-modifying enzyme(again adding an amino group, in this case provided by TrpG)and lyase (1, 56). Recently, a TrpE-like salicylate synthase,Irp9, which is capable of both ring isomerization and pyruvatelyase activity, from Y. enterocolitica has been characterized(30, 41).

Here we describe the crystal structure of MbtI, the geneproduct of Rv2386c from M. tuberculosis, the presumed iso-chorismate synthase that catalyzes the first step in mycobactinbiosynthesis. We present both structural and biochemical evi-dence to demonstrate that MbtI does not function as an iso-chorismate synthase like PchA but instead as a salicylate synthaselike Irp9, without the need for the separate isochorismate-pyru-vate lyase (PchB) required in Pseudomonas (17). The structureof MbtI shows that the catalytic apparatus it shares with the

FIG. 1. Chemical structure of mycobactin T, the siderophore of M. tuberculosis, annotated with the biosynthetic origins of its constituentcomponents. In water-soluble variants, the acyl chain on the central modified lysine is 2 to 8 carbons long; in cell-wall associated variants, it canbe up to 20 carbons in length. The drawing is based on data from reference 13.

6082 HARRISON ET AL. J. BACTERIOL.

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

chorismate-utilizing enzymes TrpE and PabB is essentially un-changed but that substantial variation occurs in the N-terminalregion of the protein. Distinct from TrpE and PabB, but incommon with Irp9, MbtI is found not to contain a tryptophanbinding site in its N-terminal region.

MATERIALS AND METHODS

Protein expression, purification, and crystallization. Native MbtI was ex-pressed and purified as described previously (26). Selenomethionine (SeMet)-substituted MbtI was prepared in a similar manner by expression of the constructin DL41(DE3)-CodonPlus-RP cells grown in LeMaster defined medium con-taining 25 �g/ml SeMet and appropriate antibiotics (14). Purification was carriedout as for native protein, except that all buffers were supplemented with 5 mMmercaptoethanol. Crystallization was carried out as described previously (26).Needle-shaped crystals, usually obtained as bundles of rods fused at one end,grew in 4 to 5 days by vapor diffusion from drops made by mixing equal volumesof protein solution (10 mg ml�1 in 20 mM HEPES, pH 8.0, 1% glycerol) andprecipitant solution (15% polyethylene glycol 4000, 0.2 M imidazole-malate, pH6.0). Crystals of the SeMet-substituted protein grew under similar conditions butrequired a higher protein concentration (20 mg ml�1).

Data collection and processing. For data collection, crystals were soaked in acryoprotectant comprising reservoir solution (15% polyethylene glycol 4000, 0.2M imidazole-malate, pH 6.0) supplemented with 20% (vol/vol) glycerol and werethen flash cooled at 113 K. Native data to 1.8-Å resolution were collected onbeamline 14-4 at the European Synchrotron Radiation Facility (ESRF),Grenoble, France, with an ADSC Quantum4 detector at a wavelength of 0.9393Å. For the SeMet-substituted crystals, data were collected from a single crystal attwo wavelengths (peak and inflection) on beamline 5.0.2 of the Advanced LightSource (Lawrence Berkeley Laboratory) by use of an ADSC Q315 charge-coupled-device detector. The data were indexed and integrated using MOSFLM(7). Both native and SeMet-substituted crystals proved to be orthorhombic, spacegroup P212121, with unit cell dimensions as follows: a � 51.82 Å, b � 163.36 Å,c � 194.93 Å (native) and a � 51.85 Å, b � 163.60 Å, c � 194.54 Å (SeMet). Thenative data set was 99.5% complete to 1.8-Å resolution, with an overall Rmerge of8.6% on intensities. Significant radiation decay occurred for the SeMet-substi-tuted crystal, limiting data collection to two wavelengths. Full data collectionstatistics are given in Table 1.

Structure determination and refinement. Selenium positions were located andrefined with autoSHARP (52) by using all data from 60 to 2.6 Å. All of the 24expected selenium atoms in the asymmetric unit were located, confirming thepresence of four molecules in the asymmetric unit. Real space density modifi-cation by solvent flattening and histogram matching was carried out using DM

(9). Initial electron density maps were of sufficiently high quality to allow auto-building using ARP/wARP (42). Initial autobuilding was able to place side chainsfor 362 residues out of a possible 450 from one molecule in the asymmetric unit.This model was then used for molecular replacement into the high-resolutionnative data set collected previously (26) by use of MOLREP (51). The resultingSigmaA-weighted electron density map showed clear density for many missingside chains and some loops. Several rounds of manual building using O (28) werethen interspersed with refinement with CNS (2), using all data to 1.8 Å. Refine-ment strategy was judged by monitoring the free R factor (Rfree) calculated from5% of reflections. Initial simulated annealing was carried out with noncrystallo-graphic symmetry restraints, bulk solvent correction, and an overall anisotropicB factor in place. In later stages, the noncrystallographic restraints were released,and individual, restrained atomic B factors were refined. Water molecules wereadded once the R factor dropped below 0.25. Glycerol, pyruvate, and imidazolemolecules were modeled late in the refinement process, into peaks of �3� inFo�Fc maps that were too large to be water. Analysis with PROCHECK (34)showed that 93.3% of residues were in the most favored region of the Ram-achandran plot. Full details of the refined structure are shown in Table 2.

Enzyme assays. Salicylate production was monitored either by direct measure-ment using fluorescence (excitation wavelength, 305 nm; emission wavelength,410 nm) or by reversed-phase high-performance liquid chromatography (HPLC)and mass spectrometry (MS). Recombinant protein (10.2 �M) was incubated for60 min at 30°C with 1.5 mM chorismate in 20 mM HEPES, 10 mM MgCl2, pH7.5, in a total volume of 1 ml. For HPLC-MS analysis, heptafluorobutyric acid(HFBA) was added to a final concentration of 0.005% (vol/vol). A sample (5 �l)was then injected immediately onto a Zorbax analytical 3.5-�m C18 reversed-phase HPLC column (100 mm by 0.3 mm; 300-Å pore size). The flow rate was 6�l/min. Solvent A was 0.1% (vol/vol) formic acid and 0.005% (vol/vol) HFBA inwater, and solvent B was 0.1% (vol/vol) formic acid and 0.005% (vol/vol) HFBAin acetonitrile. The following gradient was used for all runs: 0 min, 2% solvent B;20 min, 20% solvent B; and 28 min, 95% solvent B. The elution of compoundswas monitored using a Q-STAR XL electrospray quadrupole mass spectrometerin negative ionization mode, using time-of-flight scanning over a range from 100to 250 m/z.

To monitor salicylate production by MbtI using 1H-nuclear magnetic reso-nance (NMR), the protein was buffer exchanged from 20 mM HEPES, 1%glycerol, pH 8.0, into 20 mM potassium phosphate buffer in D2O, pH 7.0.Salicylate and chorismate standards (2 mM in water) were added to 20 mMphosphate buffer in D2O immediately prior to use. The activity of the enzyme, asassessed by fluorimetry, was significantly reduced in phosphate buffer, and thereaction mixture (8.9 �M MbtI, 2 mM chorismate, 20 mM potassium phosphatebuffer, 10 mM MgCl2, pH 7.0) was therefore incubated for 18 h at room tem-perature. 1H-NMR spectra were collected at room temperature by use of aBruker Avance 600-MHz spectrometer equipped with a cryoprobe.

FIG. 2. Structurally characterized chorismate-utilizing pathways. Shown are the three analogous transformations of chorismate for which thereis structural information about the enzymes involved: anthranilate synthesis, p-aminobenzoate synthesis, and salicylate synthesis. ADC, 4-amino-4-deoxychorismate; ADIC, 2-amino-2-deoxyisochorismate.

VOL. 188, 2006 STRUCTURE OF MbtI FROM M. TUBERCULOSIS 6083

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

Protein structure accession number. The Protein Data Bank (PDB) accessionnumber for MbtI is 2G5F.

RESULTS

Crystal structure of MbtI. The crystal structure of MbtI wassolved by the multiwavelength anomalous diffraction methodusing crystals of the SeMet-substituted protein and was thenrefined with native data at 1.8-Å resolution to a final R factorof 0.205 and free R of 0.240. The crystal asymmetric unit wasfound to contain four MbtI molecules. The crystal packing,however, suggests a monomeric enzyme, consistent with sizeexclusion chromatography data which suggest a monomer insolution (data not shown). In the crystal asymmetric unit, thelargest buried surface between molecules is 1,200 Å2 (600 Å2

per monomer, or 3.4% of the monomer surface) between mol-ecules A and B and 1,040 Å2 between molecules A and D, ascalculated using the protein-protein interaction server (http://www.biochem.ucl.ac.uk/bsm/PP/server).

The final model conforms well to standard geometry, with93.3% of residues having their main chain torsion angles in themost favored regions of the Ramachandran plot, as defined byPROCHECK. Molecules A, C, and D each comprise 434 res-idues (residues 15 to 449), with only the N-terminal residues 1to 14 and C-terminal residue 450 having no interpretable elec-tron density. Molecule B is additionally lacking density forresidues 38 to 41 and 76. Overall, the four protomers are verysimilar: molecules A and B superpose with an overall rootmean square (RMS) difference in C� positions of 0.59 Å, andmolecules C and D superpose with an RMS difference of 0.48Å. Molecules A and B are fractionally more similar to eachother than they are to molecules C and D (RMS differences:A/C, 0.93 Å; A/D, 0.88 Å; B/C, 0.97 Å; B/D, 1.01 Å), but thereare no significant structural changes between the four mono-mers apart from localized differences in the active site, asdescribed in more detail below. Full details of the final modelare given in Table 2.

Molecular structure. The polypeptide of 450 residues formsone large single domain, with a fold that is basically the sameas that previously described for the chorismate-utilizing en-zymes Irp9, TrpE, and PabB. This fold can be described interms of two �/� subdomains, each comprising a large antipa-rallel �-sheet with helices packed against it. The two �-sheets,one of 10 strands (subdomain I) and the other of 11 strands(subdomain II), pack approximately orthogonally across eachother in the center of the molecule (Fig. 3), burying hydropho-bic side chains between them. Helices pack on the outside ofthe �-sheets, with five helices packed against the subdomain I�-sheet and six packed against the subdomain II �-sheet.

Outside the region where the two �-sheets cross, a deep cleftis formed between regions of the two subdomains. Severaladventitiously bound ligands are found in this cleft, which byanalogy with the substrate-bound structures of Irp9 (29) andTrpE (50) is identified as the active site. Clear electron density

TABLE 1. Data collection and phasing statistics

Parameter

Value forb:

Native protein SeMetprotein

Data collectionWavelength (Å) 0.9393 0.9796 0.9797Resolution range (Å) 46.6–1.8 (1.9–1.8) 60.0–2.6 (2.74–2.6) 60.0–2.6 (2.74–2.6)No. of total reflections 586,276 362,879 365,449No. of unique reflections 153,449 51,122 52,360Completeness (%) 99.5 (97.6) 100 (100) 99.4 (98.4)Multiplicity 3.8 (3.4) 7.1 (7.0) 7.0 (6.6)Avg I/�(I) 4.8 (1.8) 5.8 (2.5) 9.2 (2.9)Rmerge

a 0.086 (0.400) 0.087 (0.240) 0.061 (0.240)

Phasing statisticsPhasing power 0.989Rcullis

c 0.817FOMd (SHARP) 0.22FOM (after solvent flattening) 0.77FOM (after DM) 0.79

a Rmerge � |I�I�|/I (where I is the intensity of an individual reflection and I� is the mean intensity of all reflections).b Values in parentheses are for the outer resolution shell.c Rcullis � �E/�D, where E is residual lack of closure error and D is anomalous difference.d FOM, figure of merit.

TABLE 2. Refinement statistics

Parameter Value

Resolution range (Å)......................................................... 50.0–1.8No. of reflections (working/test) ......................................138,031/15,321Rcryst/Rfree ............................................................................ 0.205/0.240No. of atoms (nonhydrogen)

Protein ............................................................................. 13,115Water ............................................................................... 994Glycerol/pyruvate/imidazole.......................................... 36/12/20

RMS deviations from idealityBonds (Å)........................................................................ 0.005Angles (°) ........................................................................ 1.3

Ramachandran plot analysis (% residues)Most favored region....................................................... 93.3Additional allowed regions ........................................... 6.5Generously allowed regions .......................................... 0.2Disallowed region........................................................... 0

6084 HARRISON ET AL. J. BACTERIOL.

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

found adjacent to Arg405, from the �19-�20 loop, is modeledas a molecule of pyruvate in molecules A and B. Less-well-defined electron density in the same general area of the struc-ture in molecules C and D is interpreted as glycerol, as dis-cussed in more detail below. A further glycerol molecule isfound adjacent to His334 from the �16-�17 loop in each of thefour MbtI molecules. On the external surface, and without anyobvious functional significance, planar electron density is in-terpreted as an imidazole molecule hydrogen bonded toGlu307 of each molecule.

Comparison with related enzymes. A BLAST search of thecurrent nonredundant sequence database with the MbtI se-quence reveals many homologous sequences, reflecting thewidespread occurrence of enzymes that act on chorismate ina number of metabolic pathways. The most similar homo-logs in sequence terms identified include anthranilate synthase(TrpE) from a number of bacterial species, a 2,3-dihydroxy-benzoate-AMP ligase from Ralstonia solanacearum, a putativep-hydroxybenzoate synthase from Nocardia farcinica, a puta-tive salicylate synthase from Yersinia pestis, and the salicylatesynthase (Irp9) from Y. enterocolitica.

An exhaustive search of the PDB using the secondary struc-ture matching program SSM (32) shows that the closest struc-tural match to MbtI is Irp9, the salicylate synthase from Y.enterocolitica (29). For this enzyme, which shares 34% se-quence identity with MbtI, 330 residues can be superimposed

onto corresponding residues in MbtI with an RMS differencein C� atomic positions of 1.3 Å after applying a cutoff of 4.0 Å.There are small differences in terms of insertions and deletionswhen comparing MbtI to Irp9, and several loops are in con-siderably altered conformation between the two structures,notably between �4 and �4b, between �4b and �2, between �3and �6, and at the start of �4. A somewhat lower correspon-dence is found with the anthranilate synthase (TrpE) enzymesfrom S. marcescens (50) and S. enterica serovar Typhimurium(36), which share about 23% sequence identity with MbtI andfor which approximately 320 residues can be superimposedonto MbtI with an RMS difference of 1.5 Å. A striking featureis that both MbtI and Irp9 show much greater correspondencewith TrpE and PabB over their C-terminal portions. Thus,when MbtI and S. marcescens TrpE are superimposed, residues187 to 446 of MbtI match residues 245 to 510 of TrpE, withinsertions or deletions occurring at only three sites: MbtI hasone additional residue in the �10-�5 loop, one fewer in the�13-�14 loop, and six fewer in the �15-�6 loop. In contrast, inthe N-terminal portion, which includes the tryptophan regula-tory site in TrpE, major deletions occur in MbtI between �2and �3a (7 residues), �4 and �5 (19 residues), and �8 and �3a(17 residues). This pattern of structural similarity is echoed insequence similarity: the N-terminal portion of MbtI (residues1 to 186) shows sequence identity with TrpE at only 9 positions(7.5%) after structural alignment, whereas the C-terminal por-

FIG. 3. Ribbon diagram of the monomer structure of MbtI, colored blue to red from the N to the C terminus, with secondary structure elementsas defined by DSSP, and labeled to be consistent with the nomenclature used for TrpE (50). The active site cleft is highlighted with a cyan circle.

VOL. 188, 2006 STRUCTURE OF MbtI FROM M. TUBERCULOSIS 6085

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

6086 HARRISON ET AL. J. BACTERIOL.

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

tion (residues 187 to 449) shows identity at 69 positions(26.2%).

A similar relationship exists between MbtI and PabB, al-though the structural correspondence is less close: with a 4.0-Åcutoff, 304 residues match with an RMS difference in C� po-sitions of 1.8 Å. Again the C-terminal region (residues 187 to446) matches best and there are much greater differences inthe N-terminal regions. A structure-based multiple sequencealignment of MbtI, TrpE, PabB, and Irp9 is shown in Fig. 4.

Active site. The active site of MbtI, identified by comparisonto the product-bound forms of Irp9 and S. marcescens TrpE, issituated in a deep cleft between structural elements from theC-terminal portion of the molecule. The cleft takes the form ofa long groove about 12 Å in length, 10 Å deep, and 7 Å wide.One wall of this groove is formed by strand �21 and thefollowing C-terminal helix, �11 (from subdomain I), and theother by the �16-�17 loop, helix �7, and the �15-�6 loop (fromsubdomain II), while the �19-�20 and �12-�13 loops form thebottom of the cleft (Fig. 3 and 5).

In both TrpE and PabB, the �16-�17 loop (residues 387 to401 in S. marcescens TrpE and 328 to 342 in PabB) forms amobile element that can adopt a closed or open conformation,apparently depending on whether ligands are bound or not. Inour MbtI structure, this loop (residues 323 to 337) is in theopen state, in contrast to the equivalent loop (residues 310 to324) in the salicylate- and pyruvate-bound Irp9 structure,where it is in the closed state (29). The open conformation isequivalent to that seen in the tryptophan-bound form of TrpE(50), and a movement of some 5 to 7 Å at the tip of the loopwould be needed to give rise to the closed conformation foundin the Irp9 structure and previously seen in the benzoate-bound form of TrpE (50). The movement of this loop is ac-companied by a similar but slightly smaller displacement of theadjacent �14-�6 connection (Fig. 5A).

The residues that are implicated in substrate binding andcatalysis in Irp9, TrpE, and PabB are highly conserved in MbtI(Fig. 5B). The residues that coordinate the essential Mg2� ionin Irp9 and TrpE, either directly or indirectly, are all con-served: Glu294, Glu297, Glu431, and Glu434 in MbtI corre-spond to Glu281, Glu284, Glu417, and Glu420 in Irp9 and toGlu358, Glu361, Glu495, and Glu498 in TrpE. No boundMg2� is present in MbtI, however, and in its absence Glu297 isturned away to hydrogen bond to His204. The residues thatbind the salicylate in Irp9, Thr258, and His321 and the benzo-ate moiety in TrpE, Thr329, and His398 are also present inMbtI as Thr271 and His334. However, His334 is on the mobile�16-�17 loop and Thr271 on the adjacent �14 strand, so areboth swung away from the active site in this “open” form; aligand-binding role for these residues may be the major factorin loop closure. Adjacent to this part of the active site isAla269, which is conserved as alanine in the Irp9 and TrpE

enzymes but corresponds to Lys274 in PabB, where it has beenproposed to discriminate between C-2 and C-4 substitution onchorismate substrates (39) and has recently been shown to beof critical importance for catalytic activity, as it forms a cova-lent intermediate during the reaction (3, 4, 27).

At the bottom of the MbtI active site is electron density fora bound ligand, which is modeled as pyruvate in molecules Aand B. Less-well-defined electron density in a similar but notexactly equivalent position is modeled as glycerol in moleculesC and D. The electron density, B factors on refinement, anddifferences in binding mode all support these assignments. Theputative pyruvate molecules in the A and B active sites bothmake several specific interactions with surrounding sidechains; the carboxylate group forms a doubly hydrogen-bondedsalt bridge with Arg405 (O1-N 1 and O2-N 2) and additionalhydrogen bonds with Tyr385 O and the peptide NH ofGly419, and the hydroxyl group forms a hydrogen bond withLys438 (Fig. 5C). The refined B factors of the pyruvate mole-cules, 35.4 Å2 in molecule A and 25.0 Å2 in molecule B, arevery close to those for surrounding parts of the protein struc-ture. After superposition of the MbtI with the TrpE and Irp9structures, the positions of the proposed pyruvate moleculesare displaced by only 1.5 Å and 1.7 Å, respectively, whilemaking essentially the same hydrogen bonds. (In TrpE andIrp9, the equivalent arginine hydrogen bonds to O2 and O3 ofthe pyruvate via Nε and N 1.) In contrast, the putative glycerolmolecules in molecules C and D are not specifically bound,with the ligand in molecule C hydrogen bonded only to waterand that in molecule D to Thr441 O�1 and water, and haverelatively high B values after refinement—60.7 Å2 and 47.9 Å2

in molecules C and D, respectively. Arg405 adopts markedlydifferent conformations in the presence and absence of boundpyruvate, in the latter case swinging away from the active siteto make hydrogen bonds with Asn231 and the backbone oxy-gen of Thr441. However, the rest of the active site seemsunchanged in the absence of pyruvate.

Catalytic activity of MbtI. The presence of adventitiouslybound pyruvate in the structure indicated that MbtI may befunctioning as a salicylate synthase, i.e., that it has both iso-chorismate synthase (chorismate isomerase) and isochoris-mate-pyruvate lyase activities. To test whether this was thecase, the activity of the enzyme was monitored fluorometrically(Fig. 6A). When incubated with chorismate, the enzyme gaverise to a product with a fluorescent emission maximum at 410nm, which is consistent with salicylate. The reaction was abso-lutely dependent on the presence of Mg2� and appeared to bestrongly influenced by choice of buffer, with phosphate show-ing a marked slowing of reaction progress compared to Tris orHEPES. To confirm the identity of the fluorescent product assalicylate, the products of the reaction were separated by re-versed-phase HPLC, compared to standard compounds, and

FIG. 4. Structure-based multiple sequence alignment of MbtI with related enzymes of known structure. Irp9, salicylate synthase from Y.enterocolitica (PDB accession no. 2FN0/2FN1); PabB, p-aminobenzoate synthase from E. coli (PDB accession no. 1K0E/1K0G); TrpE, anthranilatesynthase from S. marcescens (PDB accession no. 1I7Q/1I7S). Structural alignment was made using sPDBv (25) and subsequently edited by hand.The figure was drawn using ESPript (23). Identical residues are shown in white type with a black background, and similar residues which areconserved in all four sequences are shown boxed. (Similar groups as defined by ESPript are as follows: HKR, polar positive; DE, polar negative;STNQ, polar neutral; AVLIM, nonpolar aliphatic; FYW, nonpolar aromatic; PG, structure breakers; and C, thiol.) Secondary structure elementsfor MbtI are shown above the alignment, labeled to be consistent with the nomenclature used for TrpE (31, 50).

VOL. 188, 2006 STRUCTURE OF MbtI FROM M. TUBERCULOSIS 6087

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

identified by electrospray mass spectrometry (Fig. 6B). Cho-rismate alone eluted with a retention time (tR) of 6.7 min, anda salicylate standard eluted with a tR of 13.2 min. The productof the MbtI reaction yielded a peak with a retention time of13.5 min. Electrospray mass spectrometric analysis indicatedthat this peak was composed of a species with an m/z of 137.0,which is consistent with salicylate and identical to the m/z ratioobtained from a standard solution of salicylate. An additionalpeak was observed on the chromatogram at a tR of 6.0 min,showing an in-source fragmentation pattern identical to that

produced by chorismate, and hence was assumed to be isocho-rismate.

In order to further analyze the reaction, 1H-NMR spectros-copy was carried out. After incubation of MbtI with choris-mate, peaks corresponding to salicylate with �H values around7.65, 7.30, and 6.80 were readily apparent (Fig. 6C). Addition-ally, peaks consistent with the presence of isochorismate, with�H values of 6.68 and 6.06, were also present, indicating thatisochorismate was likely being formed as an intermediate inthe reaction, as has been shown previously for Irp9 (30).

FIG. 5. (A) Comparison of the active site regions of MbtI and the ligand-bound conformation of TrpE, showing the shift in the �14-�17-�16sheet. MbtI is shown in green, and TrpE is shown in blue with side chains in yellow. The bound Mg2� ion is shown as a yellow sphere. In all casesexcept where noted, molecule B from the MbtI asymmetric unit is shown. (B) Overlay of the MbtI, TrpE, and Irp9 active sites. MbtI is shown ingreen, TrpE in yellow, and Irp9 in blue. (C) Pyruvate binding in the active site of MbtI. A molecule of pyruvate is shown modeled into residualdensity from a SigmaA-weighted Fo�Fc electron density map, contoured at 3�, in the active site of molecule B in the asymmetric unit. Side chainsare shown in green for residues that interact with the pyruvate, and hydrogen bonds are shown as dashed lines. This arrangement is identical inmolecule A of the asymmetric unit. The alternative conformation of Arg405 in molecules C and D in the asymmetric unit, where no bound pyruvateis observed, is shown in light blue, represented using the side chain from molecule C. (D) The tryptophan binding site of TrpE is overlaid onthe equivalent region of MbtI. The hydrogen bonding network in MbtI is shown as black dashed lines. Molecules are colored as describedfor panel A.

6088 HARRISON ET AL. J. BACTERIOL.

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

Tryptophan binding site. Although MbtI matches TrpE andPabB very closely over its entire C-terminal region and has anactive site extremely similar to that of Irp9, there are substan-tial differences to these two enzymes in their N-terminal re-gions. These have the effect that the binding site for trypto-phan, which gives feedback inhibition of TrpE and stabilizationof the PabB structure, is absent from both MbtI and Irp9.Several key substitutions in MbtI, coupled with the truncationand displacement of its �2-�3 loop, combine to abolish thetryptophan binding site (Fig. 5D). The deletion of 7 residuesfrom the �2-�3 loop and its displacement by 4 to 5 Å relativeto that in TrpE cause the C� atom of Cys50 (equivalent toSer40 in TrpE) to occupy the position filled by the tryptophanC� atom in both TrpE and PabB. In Irp9, Arg37 is the equiv-alent residue of Cys50 in MbtI and blocks the binding site in asimilar manner, hydrogen bonding to Tyr42 and Asp380. Theposition of the tryptophan indole ring in TrpE and PabB isfilled by the side chain of Tyr48 in MbtI (Tyr35 in Irp9),replacing a leucine (Leu38) that packs against the indole ringof the bound tryptophan in TrpE. The Tyr48 side chain is heldin place by a hydrogen bond to Asp399 (Asp385 in Irp9), whichalso forms a double salt bridge with Arg235. In Arg235 isanother key substitution, as it replaces Tyr292 in TrpE andTrp241 in PabB, which help form the tryptophan site. Arg223in Irp9 is equivalent to Arg235 in MbtI and forms an equiva-lent salt bridge. In both salicylate synthases, the tryptophanbinding site that is conserved among other chorismate-utilizingenzymes has clearly been ablated and replaced by a tight net-work of hydrogen-bonded residues, although the residues usedare different in MbtI than in Irp9.

Oligomerization interface. Although structurally homolo-gous, members of the group of chorismate-utilizing enzymesexemplified by MbtI, TrpE, and PabB show intriguing differ-ences in oligomerization behavior. Anthranilate synthase is aheterotetramer (stoichiometry, TrpE2:TrpG2) with the TrpEand TrpG (amidotransferase) subunits forming a tight het-erodimer, whereas PabB forms only a transient and structurallyuncharacterized complex with its equivalent amidotransferase,PabA. As there is no requirement for an amidotransferaseactivity in the conversion of chorismate to salicylate, it is un-surprising that the TrpG-binding interface of TrpE has notbeen conserved in MbtI. TrpE binds to TrpG primarily via its�7 helix, with minor contributions from the �4-�10, �21-�11,and �4b-�2 loops, the last of which is much truncated in MbtI,PabB, and Irp9. In particular, specific interactions made withTrpG by Met364, Asp367, Arg370, and Asn371 of TrpE wouldnot be possible in MbtI. Additionally, the electrostatic poten-tials of the protein surfaces in this region are rather different,especially at the C-terminal end of helix �7, which is markedlymore negatively charged in MbtI due to the presence ofGlu307, Glu308, and Asp311.

FIG. 6. Enzymatic activity of MbtI. (A) MbtI activity was moni-tored using fluorometric detection at an emission wavelength of 410nm in the presence (�) and absence (�) of Mg2�. (B) Extracted ionchromatograms of MbtI reaction products, sampled over an m/z rangefrom 137.0 to 137.1, compared to chromatograms of chorismate andsalicylate incubated under the same conditions without enzyme. Chro-matograms are all normalized to the highest peak. Chorismate yields afragment with an m/z of 137 due to in-source fragmentation and alsoyields other fragments at an m/z of 207 and an m/z of 225 (traces not

shown for clarity). (C) 1H-NMR spectroscopic analysis of salicylateproduction by MbtI, compared to spectra of chorismate and salicylatealone.

VOL. 188, 2006 STRUCTURE OF MbtI FROM M. TUBERCULOSIS 6089

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

DISCUSSION

The production of the salicylate-based siderophore myco-bactin appears to be a critical component of virulence in M.tuberculosis, as its ability to scavenge iron while in the macro-phage phagosome sets it apart from nonpathogenic mycobac-terial species (53). The TrpE homologue MbtI has been iden-tified by genetic organization and gene regulation to be a likelycandidate for salicylate production in M. tuberculosis. Here wepresent structural and in vitro biochemical evidence that MbtIis a salicylate synthase, capable of both ring isomerization andpyruvate lyase activity. A bound molecule of pyruvate identi-fied in the active site indicated that the enzyme in the formcrystallized had presumably metabolized chorismate from theexpression host and that pyruvate lyase activity was likely tohave occurred. This activity was confirmed by using fluorime-try, NMR, and HPLC-MS to detect salicylate production invitro. This identifies salicylate biosynthesis in M. tuberculosis asbeing a single-step process like that seen with Y. enterocolitica(29, 30, 41) rather than the two-step process characterized forP. aeruginosa, as had been speculated previously (18, 44). Thisconfirmation that M. tuberculosis carries out a single-step sa-licylate synthesis is consistent with the original biochemicalevidence for a bacterial salicylate biosynthetic pathway, whichcame from the identification of shikimate-derived salicylate inM. smegmatis, where the presumed two components involvedin its production were unable to be separated (35).

Although several structures are now available for choris-mate-utilizing enzymes, many questions remain about themechanisms involved in the reactions they catalyze. One un-answered question is why some of the enzymes (PabB andPchA) require an additional lyase (PabC and PchB, respec-tively) to remove the pyruvate moiety from the reaction inter-mediate, while other enzymes (MbtI, TrpE, and Irp9) are ableto catalyze this step without assistance. The pyruvate lyaseenzymes PabC and PchB are themselves not related and areproposed to work via quite distinct mechanisms: the former isa pyridoxal phosphate-containing enzyme related to aminoacid aminotransferases (24), while the latter shows weak cho-rismate mutase activity and is related to AroQ-type chorismatemutases (17). It has been suggested that the presence ofLys274 in PabB may both influence the specificity of the reac-tion and predict the need for a separate lyase (39). Arguingagainst this being so, PchA has an alanine residue in the equiv-alent position, in common with the other lyase-competent en-zymes TrpE, MbtI, and Irp9. Another a priori possibility is thatsubstitutions at position 2 of the chorismate ring, as seen inanthranilate and salicylate synthases, simply form a less chem-ically stable intermediate, which itself rearranges to form thereaction products. The fact that PchB is required for lyaseactivity in P. aeruginosa argues against this also, but it is inter-esting in this context to note that recent work with PchB hasindicated the likelihood of the reaction proceeding via an un-usual pericyclic mechanism (11, 33) rather than by the previ-ously proposed general base proton abstraction and elimina-tion reaction (54). Site-directed mutational analysis supportsthe idea that the same pericyclic mechanism is used in Irp9(29), and by extension this is likely to be the case in MbtI also.From a structural perspective and in the absence of a structureof PchA, it is difficult to make any definitive comment on what

the critical determinant of lyase capability might be, as all ofthe critical residues thought to be involved in the catalyticprocess are conserved between the lyase-competent MbtI andIrp9 and the lyase-dependent PchA.

The proposed pericyclic mechanism for lyase activity doesnot require a general base but does require a positivelycharged residue to stabilize a negative charge on the etheroxygen of the isochorismate intermediate (33). In the structureof MbtI, Lys438 is hydrogen bonded to the ketone (O3) oxygenof the bound pyruvate (Fig. 5C), which would have formed theether bridge in the isochorismate intermediate. This residue isabsolutely conserved, though it does not make any direct in-teractions with substrate or products in any of the other knownstructures. Its position, and the observation that it is able tomake a hydrogen bond to the oxygen in question, makes it aplausible candidate for contributing the stabilizing positivecharge required by this mechanism.

Although MbtI appears to be a monomeric enzyme in solu-tion, in the crystal structure it makes an intersubunit contactthat is equivalent to the dimerization interface of Irp9 and ofS. marcescens TrpE, involving the helices �6 and �7 and betastrand �16. As active site residues are contributed from �7 andfrom the adjacent �17 strand, it is possible that enzyme activitymay be influenced by multimeric assembly. Although the struc-ture of the enzyme has been described in terms of two subdo-mains, each constituting a large �-sheet, it could equally wellbe described as being composed of two domains, a smaller onemade up from helices �6 and �7 and �-strands �14, �16, and�17 (Fig. 3) and a larger domain made up of the rest of theprotein. There is inherent, functionally significant flexibilityapparent in the region, as seen here with MbtI and also withPabB, where the �14-�16-�17 sheet can place the active siteinto open and closed conformations (Fig. 5A).

Given the critical importance of iron acquisition for patho-genicity in M. tuberculosis, siderophore biosynthesis representsa clearly validated target for the development of new antitu-berculosis drugs. Salicyl-AMS, an inhibitor of MbtA, a bifunc-tional salicyl-AMP ligase/salicyl-S-ArCP synthetase whichtransfers salicylate onto the aroyl carrier protein (ArCP), hasrecently been synthesized and showed marked in vitro growthinhibition of M. tuberculosis, with 50% inhibitory concentrationvalues in the �M range (16). The structure of MbtI presentedhere represents the starting point for the rational design ofsmall molecule inhibitors of salicylate synthesis in M. tubercu-losis. An initial series of chorismate-based competitive inhibi-tors of Irp9 has recently been described (40), the most potentof which had a Ki of �20 �M, which may provide a guide forthe synthesis of specific inhibitors of MbtI.

ACKNOWLEDGMENTS

We thank Didier Nurizzo (European Synchrotron Radiation Facil-ity) for native X-ray data collection and Michael Walker (Departmentof Chemistry, University of Auckland) for NMR data collection.

This work was funded by the Health Research Council of NewZealand and a New Economy Research Fund grant from the NewZealand Foundation for Research, Science and Technology.

REFERENCES

1. Bauerle, R., J. Hess, and S. French. 1987. Anthranilate synthase-anthranilatephosphoribosyltransferase complex and subunits of Salmonella typhimurium.Methods Enzymol. 142:366–386.

2. Brunger, A. T., P. D. Adams, G. M. Clore, W. L. DeLano, P. Gros, R. W.

6090 HARRISON ET AL. J. BACTERIOL.

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

Grosse-Kunstleve, J. S. Jiang, J. Kuszewski, M. Nilges, N. S. Pannu, R. J.Read, L. M. Rice, T. Simonson, and G. L. Warren. 1998. Crystallography &NMR system: a new software suite for macromolecular structure determi-nation. Acta Crystallogr. D 54:905–921.

3. Bulloch, E. M. M., and C. Abell. 2005. Detection of covalent intermediatesformed in the reaction of 4-amino-4-deoxychorismate synthase. Chembio-chem 6:832–834.

4. Bulloch, E. M. M., M. A. Jones, E. J. Parker, A. P. Osborne, E. Stephens,G. M. Davies, J. R. Coggins, and C. Abell. 2004. Identification of 4-amino-4-deoxychorismate synthase as the molecular target for the antimicrobialaction of (6S)-6-fluoroshikimate. J. Am. Chem. Soc. 126:9912–9913.

5. Clemens, D. L., and M. A. Horwitz. 1996. The Mycobacterium tuberculosisphagosome interacts with early endosomes and is accessible to exogenouslyadministered transferrin. J. Exp. Med. 184:1349–1355.

6. Cole, S. T., R. Brosch, J. Parkhill, T. Garnier, C. Churcher, D. Harris, S. V.Gordon, K. Eiglmeier, S. Gas, C. E. Barry III, 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, and B. G.Barrell. 1998. Deciphering the biology of Mycobacterium tuberculosis fromthe complete genome sequence. Nature 393:537–544.

7. Collaborative Computational Project Number 4. 1994. The CCP4 suite:programs for protein crystallography. Acta Crystallogr. D 50:760–763.

8. Corbett, E. L., C. J. Watt, N. Walker, D. Maher, B. G. Williams, M. C.Raviglione, and C. Dye. 2003. The growing burden of tuberculosis. Arch.Intern. Med. 163:1009–1021.

9. Cowtan, K., and P. Main. 1998. Miscellaneous algorithms for density modifica-tion. Acta Crystallogr. D54:487–493.

10. Crosa, J. H., and C. T. Walsh. 2002. Genetics and assembly line enzymology ofsiderophore biosynthesis in bacteria. Microbiol. Mol. Biol. Rev. 66:223–249.

11. DeClue, M. S., K. K. Baldridge, D. E. Kunzler, P. Kast, and D. Hilvert. 2005.Isochorismate pyruvate lyase: a pericyclic reaction mechanism? J. Am.Chem. Soc. 127:15002–15003.

12. De Voss, J. J., K. Rutter, B. G. Schroeder, H. Su, Y. Zhu, and C. E. Barry.2000. The salicylate-derived mycobactin siderophores of Mycobacteriumtuberculosis are essential for growth in macrophages. Proc. Natl. Acad.Sci. USA 97:1252–1257.

13. De Voss, J. J., K. Rutter, B. G. Schroeder, and C. E. Barry. 1999. Ironacquisition and metabolism by mycobacteria. J. Bacteriol. 181:4443–4451.

14. Doublie, S. 1997. Preparation of selenomethionyl proteins for phase deter-mination. Methods Enzymol. 276:523–530.

15. Espinal, M. A. 2003. The global situation of MDR-TB. Tuberculosis 83:44–51.16. Ferreras, J. A., J.-S. Ryu, F. D. Lello, D. S. Tan, and L. E. N. Quadri. 2005.

Small-molecule inhibition of siderophore biosynthesis in Mycobacteriumtuberculosis and Yersinia pestis. Nat. Chem. Biol. 1:29–32.

17. Gaille, C., P. Kast, and D. Haas. 2002. Salicylate biosynthesis in Pseudomonasaeruginosa—purification and characterization of PchB, a novel bifunctional en-zyme displaying isochorismate pyruvate-lyase and chorismate mutase activities.J. Biol. Chem. 277:21768–21775.

18. Gaille, C., C. Reimmann, and D. Haas. 2003. Isochorismate synthase (PchA),the first and rate-limiting enzyme in salicylate biosynthesis of Pseudomonasaeruginosa. J. Biol. Chem. 278:16893–16898.

19. Garcia-del Portillo, F., J. W. Foster, M. E. Maguire, and B. B. Finlay. 1992.Characterization of the micro-environment of Salmonella typhimurium-con-taining vacuoles within MDCK epithelial cells. Mol. Microbiol. 6:3289–3297.

20. Gobin, J., and M. A. Horwitz. 1996. Exochelins of Mycobacterium tuberculosisremove iron from human iron-binding proteins and donate iron to myco-bactins in the M. tuberculosis cell wall. J. Exp. Med. 183:1527–1532.

21. Gold, B., G. M. Rodriguez, S. A. E. Marras, M. Pentecost, and I. Smith. 2001.The Mycobacterium tuberculosis IdeR is a dual functional regulator thatcontrols transcription of genes involved in iron acquisition, iron storage andsurvival in the macrophages. Mol. Microbiol. 42:851–865.

22. Gomez, J. E., and J. D. McKinney. 2004. M. tuberculosis persistence, latencyand drug tolerance. Tuberculosis 84:29–44.

23. Gouet, P., X. Robert, and E. Courcelle. 2003. ESPript/ENDscript: extractingand rendering sequence and 3D information from atomic structures of pro-teins. Nucleic Acids Res. 31:3320–3323.

24. Green, J. M., W. K. Merkel, and B. P. Nichols. 1992. Characterization andsequence of Escherichia coli pabC, the gene encoding aminodeoxychorismatelyase, a pyridoxal phosphate-containing enzyme. J. Bacteriol. 174:5317–5323.

25. Guex, N., and M. C. Peitsch. 1997. SWISS-MODEL and the Swiss-PdbViewer:an environment for comparative protein modeling. Electrophoresis 18:2714–2723.

26. Harrison, A. J., R. J. Ramsay, E. N. Baker, and J. S. Lott. 2005. Crystalli-zation and preliminary X-ray crystallographic analysis of MbtI, a proteinessential for siderophore biosynthesis in Mycobacterium tuberculosis. ActaCrystallogr. F 61:121–123.

27. He, Z., and M. D. Toney. 2006. Direct detection and kinetic analysis ofcovalent intermediate formation in the 4-amino-4-deoxychorismate synthasecatalyzed reaction. Biochemistry 45:5019–5028.

28. Jones, T. A., J. Y. Zou, S. W. Cowan, and M. Kjeldgaard. 1991. Improvedmethods for building protein models in electron density maps and the loca-tion of errors in these models. Acta Crystallogr. A 47:110–119.

29. Kerbarh, O., D. Y. Chirgadze, T. L. Blundell, and C. Abell. 2006. Crystalstructures of Yersinia enterocolitica salicylate synthase and its complex withthe reaction products salicylate and pyruvate. J. Mol. Biol. 357:524–534.

30. Kerbarh, O., A. Ciulli, N. I. Howard, and C. Abell. 2005. Salicylate biosynthesis:overexpression, purification, and characterization of Irp9, a bifunctional salicy-late synthase from Yersinia enterocolitica. J. Bacteriol. 187:5061–5066.

31. Knochel, T., A. Ivens, G. Hester, A. Gonzalez, R. Bauerle, M. Wilmanns, R.Kirschner, and J. N. Jansonius. 1999. The crystal structure of anthranilatesynthase from Sulfolobus solfataricus: functional implications. Proc. Natl.Acad. Sci. USA 96:9479–9484.

32. Krissinel, E., and K. Henrick. 2004. Secondary-structure matching (SSM), anew tool for fast protein structure alignment in three dimensions. ActaCrystallogr. D 60:2256–2268.

33. Kunzler, D. E., S. Sasso, M. Gamper, D. Hilvert, and P. Kast. 2005. Mech-anistic insights into the isochorismate pyruvate lyase activity of the catalyt-ically promiscuous PchB from combinatorial mutagenesis and selection.J. Biol. Chem. 280:32827–32834.

34. Laskowski, R. A., M. W. MacArthur, D. S. Moss, and J. M. Thornton. 1993.PROCHECK: a program to check the stereochemical quality of proteinstructures. J. Appl. Crystallogr. 26:283–291.

35. Marshall, B. J., and C. Ratledge. 1972. Salicylic acid biosynthesis and itscontrol in Mycobacterium smegmatis. Biochim. Biophys. Acta 264:106–116.

36. Morollo, A. A., and M. J. Eck. 2001. Structure of the cooperative allostericanthranilate synthase from Salmonella typhimurium. Nat. Struct. Biol. 8:243–247.

37. Nichols, B. P., A. N. Seibold, and S. Z. Doktor. 1989. para-Aminobenzoatesynthesis from chorismate occurs in two steps. J. Biol. Chem. 264:8597–8601.

38. O’Regan, A., and M. Joyce-Brady. 2001. Latent tuberculosis may persist forover 40 years. BMJ 323:635.

39. Parsons, J. F., P. Y. Jensen, A. S. Pachikara, A. J. Howard, E. Eisenstein,and J. E. Ladner. 2002. Structure of Escherichia coli aminodeoxychorismatesynthase: architectural conservation and diversity in chorismate-utilizing en-zymes. Biochemistry 41:2198–2208.

40. Payne, R. J., O. Kerbarh, R. N. Miguel, A. D. Abell, and C. Abell. 2005.Inhibition studies on salicylate synthase. Org. Biomol. Chem. 3:1825–1827.

41. Pelludat, C., D. Brem, and J. Heesemann. 2003. Irp9, encoded by the high-pathogenicity island of Yersinia enterocolitica, is able to convert chorismateinto salicylate, the precursor of the siderophore yersiniabactin. J. Bacteriol.185:5648–5653.

42. Perrakis, A., R. Morris, and V. S. Lamzin. 1999. Automated protein modelbuilding combined with iterative structure refinement. Nat. Struct. Biol.6:458–463.

43. Pym, A. S., and S. T. Cole. 1999. Post DOTS, post genomics: the next centuryof tuberculosis control. Lancet 353:1004–1005.

44. Quadri, L. E., J. Sello, T. A. Keating, P. H. Weinreb, and C. T. Walsh. 1998.Identification of a Mycobacterium tuberculosis gene cluster encoding thebiosynthetic genes for assembly of the virulence-conferring siderophore my-cobactin. Chem. Biol. 5:631–645.

45. Ratledge, C. 2004. Iron, mycobacteria and tuberculosis. Tuberculosis 84:110–130.

46. Rodriguez, G. M., and I. Smith. 2003. Mechanisms of iron regulation in myco-bacteria: role in physiology and virulence. Mol. Microbiol. 47:1485–1494.

47. Rodriguez, G. M., M. I. Voskuil, B. Gold, G. K. Schoolnik, and I. Smith.2002. ideR, an essential gene in Mycobacterium tuberculosis: role of IdeR iniron-dependent gene expression, iron metabolism, and oxidative stress re-sponse. Infect. Immun. 70:3371–3381.

48. Sassetti, C. M., D. H. Boyd, and E. J. Rubin. 2003. Genes required formycobacterial growth defined by high density mutagenesis. Mol. Microbiol.48:77–84.

49. Schaible, U. E., and S. H. E. Kaufmann. 2004. Iron and microbial infection.Nat. Rev. Microbiol. 2:946–953.

50. Spraggon, G., C. Kim, X. Nguyen-Huu, M.-C. Yee, C. Yanofsky, and S. E.Mills. 2001. The structures of anthranilate synthase of Serratia marcescenscrystallized in the presence of (i) its substrates, chorismate and glutamine,and a product, glutamate, and (ii) its end-product inhibitor, L-tryptophan.Proc. Natl. Acad. Sci. USA 98:6021–6026.

51. Vagin, A., and A. Teplyakov. 1997. MOLREP: an automated program formolecular replacement. J. Appl. Crystallogr. 30:1022–1025.

52. Vonrhein, C., E. Blanc, P. Roversi, and G. Bricogne. 2005. Automatedstructure solution with autoSHARP, p. 1–24. In S. Doublie (ed.), Crystallo-graphic methods. Humana Press, Totowa, N.J.

53. Wagner, D., M. Maser, B. Lai, Z. Cai, C. E. Barry, K. Hoener zu Bentrup,D. G. Russell, and L. E. Bermudez. 2005. Elemental analysis of Mycobacte-rium avium-, Mycobacterium tuberculosis-, and Mycobacterium smegmatis-containing phagosomes indicates pathogen-induced microenvironmentswithin the host cell’s endosomal system. J. Immunol. 174:1491–1500.

54. Walsh, C. T., J. Liu, F. Rusnak, and M. Sakaitani. 1990. Molecular studieson enzymes in chorismate metabolism and the enterobactin biosyntheticpathway. Chem. Rev. 90:1105–1129.

55. Wandersman, C., and P. Delepelaire. 2004. Bacterial iron sources: fromsiderophores to hemophores. Annu. Rev. Microbiol. 58:611–647.

56. Zalkin, H. 1985. Anthranilate synthase. Methods Enzymol. 113:287–292.

VOL. 188, 2006 STRUCTURE OF MbtI FROM M. TUBERCULOSIS 6091

on March 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from