Characterization of the secreted chorismate mutase from the pathogen Mycobacterium tuberculosis Severin Sasso, Chandra Ramakrishnan, Marianne Gamper, Donald Hilvert and Peter Kast Laboratorium fu ¨ r Organische Chemie, Swiss Federal Institute of Technology, Zu ¨ rich (ETH), Switzerland The intracellular shikimate pathway is essential in bac- teria, fungi, algae and plants for the synthesis of aro- matic compounds [1], but is absent from mammals and thus represents a promising target for antimicrobial or antifungal agents and herbicides. The branch point intermediate of the shikimate pathway is chorismate, and its partitioning towards the individual aromatic products is controlled by the activities of several chor- ismate-metabolizing enzymes. One of these is choris- mate mutase (CM; EC 5.4.99.5), which catalyzes the Claisen rearrangement of chorismate to prephenate, the committed step [2] in the biosynthesis of tyrosine and phenylalanine (Fig. 1A) [1]. Interestingly, two totally unrelated protein scaffolds have evolved to carry out the CM reaction with similar efficiencies ([3,4], Fig. 1B). Enzymes of the relatively rare AroH class, including the monofunctional CMs from Bacillus subtilis (BsCM) and Thermus thermophi- lus have a trimeric pseudo a ⁄ b barrel structure [5–7]. In contrast, proteins of the a-helical AroQ class, repre- sented by the structurally characterized CM domain of the Escherichia coli chorismate mutase-prephenate dehydratase (EcCM [8]), and the CM from the yeast Saccharomyces cerevisiae [9,10], are considerably more abundant. EcCM, the prototype for an AroQ class member, is an intertwined dimer consisting of two subunits of three a-helices each ([8], Fig. 1B). The S. cerevisiae CM is a more elaborate variant of the basic AroQ fold. It is also dimeric, but each subunit, which is believed to have arisen from a duplicated Keywords chorismate mutase; Mycobacterium tuberculosis; pathogenesis; shikimate pathway; signal sequence Correspondence P. Kast, Laboratorium fu ¨ r Organische Chemie, Swiss Federal Institute of Technology, ETH Ho ¨ nggerberg – HCI F333, CH-8093 Zu ¨ rich, Switzerland Fax: +41 1633 1326 Tel: +41 1632 2908 E-mail: kast@org.chem.ethz.ch (Received 12 October 2004, revised 4 November 2004, accepted 12 November 2004) doi:10.1111/j.1742-4658.2004.04478.x The gene encompassing ORF Rv1885c with weak sequence similarity to AroQ chorismate mutases (CMs) was cloned from the genome of Mycobac- terium tuberculosis and expressed in Escherichia coli. The gene product (*MtCM) complements a CM-deficient E. coli strain, but only if produced without the predicted N-terminal signal sequence typical of M. tuberculosis. The mature *MtCM, which was purified by exploiting its resistance to irre- versible thermal denaturation, possesses high CM activity in vitro. The enzyme follows simple Michaelis–Menten kinetics, having a k cat of 50 s )1 and a K m of 180 lm (at 30 °C and pH 7.5). *MtCM was shown to be a dimer by analytical ultracentrifugation and size-exclusion chromatography. Secondary-structure prediction and CD spectroscopy confirmed that *MtCM is a member of the all-a-helical AroQ class of CMs, but it seems to have a topologically rearranged AroQ fold. Because CMs are normally intracellular metabolic enzymes required for the biosynthesis of phenyl- alanine and tyrosine, the existence of an exported CM in Gram-positive M. tuberculosis is puzzling. The observation that homologs of *MtCM with a predicted export sequence are generally only present in parasitic or pathogenic organisms suggests that secreted CMs may have evolved to par- ticipate in some aspect of parasitism or pathogenesis yet to be unraveled. Abbreviations CM, chorismate mutase; *MtCM, secreted Mycobacterium tuberculosis CM; BsCM, Bacillus subtilis CM; EcCM, CM domain of the bifunctional Escherichia coli chorismate mutase-prephenate dehydratase; IPTG, isopropyl thio-b- D-galactoside. FEBS Journal 272 (2005) 375–389 ª 2004 FEBS 375 primordial AroQ gene of the EcCM type, is made up of 12 a-helices. Each polypeptide forms a catalytic domain, which superimposes closely on the corres- ponding EcCM structure, and an additional, divergent regulatory domain for the binding of the allosteric effectors tyrosine or tryptophan at the interface of the dimer [9,10]. It is noteworthy that the active sites of AroQ and AroH CMs are similarly functionalized, indicating convergent evolution [3,11]. Among the AroQ CMs is a subgroup, dubbed *AroQ by Jensen and coworkers [12], whose members apparently are exported from the cytoplasm [4,12,13]. So far, *AroQ proteins have been isolated from Gram- negative bacteria, such as Erwinia herbicola, Salmonella typhimurium and Pseudomonas aeruginosa [12,13], where they are targeted to another subcellular com- partment, the periplasmic space, which surrounds the cytoplasm but features a more oxidizing milieu [14]. As the classical metabolic routes to tyrosine and phe- nylalanine are entirely cytoplasmic in bacteria, export of a CM does not seem to make sense. Furthermore, secreted functional *AroQ homologs have been discov- ered in nematodes, which do not even possess a shiki- mate pathway [15–17]. Although the role of exported CMs is still obscure, it is striking that the presence of an *AroQ protein in an organism correlates with its pathogenicity [4,12,13,15–19]. With the availability of the genomic sequence, Mycobacterium tuberculosis has become a model organism for pathogenic bacteria [20,21], and great efforts are currently being made to characterize its proteome in detail [22–24]. In this work, we report the cloning of an M. tuberculosis gene the product of which is homologous to *AroQ CMs. Expression in B A Fig. 1. Chorismate mutase. (A) Biosynthesis of aromatic compounds via the shikimate pathway. Chorismate mutase (CM) catalyzes the com- mitted step of the branch towards phenylalanine and tyrosine. The information next to the arrows corresponding to enzymatic steps refers to already assigned genes in the M. tuberculosis genome [20] (http://genolist.pasteur.fr/TubercuList, see Discussion). (B) Ribbon diagrams of EcCM [8] and BsCM [5,6], the prototypic enzymes of the AroQ and AroH class, respectively, complexed with transition state analog 1 (as ball and stick model). The atomic coordinates for the crystal structures of EcCM and BsCM are available in the Research Collaboratory for Structural Bioinformatics Protein Databank under PDB numbers 1ECM and 2CHT, respectively. Secreted chorismate mutase from M. tuberculosis S. Sasso et al. 376 FEBS Journal 272 (2005) 375–389 ª 2004 FEBS E. coli demonstrates that the protein is indeed a CM, which is exported from the cytoplasm. As M. tuber- culosis is Gram-positive and thus lacks a periplasmic space, this CM is the first example of an *AroQ pro- tein that may be secreted directly into the surrounding medium and away from the bacterial cell. Further- more, this study represents the most rigorous struc- tural and functional characterization to date of a member of a topological subclass of AroQ CMs, which may be involved in the etiology of diseases such as tuberculosis [25] which claim many millions of lives every year. Results Cloning of an aroQ homolog from the M. tuberculosis genome Homology searches for aroQ genes in GenBank [26] revealed that M. tuberculosis possesses a gene (ORF Rv1885c in strain H37Rv, accession number CAB10064 in DDBJ⁄ EMBL⁄ GenBank [20]), encoding a putative protein with high similarity to the exported *AroQ CMs. Analysis of the primary sequence of the encoded protein (termed *MtCM) using the neural network pro- gram SignalP (http://www.cbs.dtu.dk/services/SignalP/ [27,28]), predicted a cleavable 33-amino-acid export sig- nal peptide at the N-terminus. We have cloned the gene (subsequently referred to as *aroQ) from the chromoso- mal DNA of M. tuberculosis and inserted it into several different plasmid constructs for in vivo and in vitro studies. Plasmids pKTU1-HCW and pKTU2-HNW carry the entire *aroQ gene, including the sequence for the presumed leader peptide, and in addition specify either a C-terminal or an N-terminal His tag, respect- ively. Plasmids pKTU3-HCT and pKTU3-HT encode a leaderless version of *MtCM corresponding to the mature form with and without a C-terminal His tag. Genetic complementation To probe whether the product of the ORF Rv1885c has CM activity in vivo, the CM-deficient E. coli KA12 ⁄ pKIMP-UAUC was provided with the *aroQ plasmids described above. KA12 ⁄ pKIMP-UAUC grows on M9c ⁄ S+F minimal medium agar plates only if Tyr is added exogenously, or if the strain receives a functional and expressed CM gene [11]. Table 1 shows that transformants that produce *MtCM bearing a sig- nal sequence do not grow in the absence of Tyr. This suggests that post-translational export of (unfolded) full-length *MtCM is very efficient [29] or that unproc- essed protein present in the cytoplasm is not very act- ive. In contrast, leaderless *MtCM complemented the CM deficiency very well, provided that the sal promo- ter was switched on by salicylate. Cells producing the leaderless but untagged protein (encoded on pKTU3- HT) grew as well as those transformed with pKTU3- HCT (data not shown). Localization of *MtCM in subcellular compart- ments and determination of the signal sequence- processing site To obtain further experimental evidence for the func- tionality of the predicted N-terminal signal sequence, the distribution of CM activity in subcellular compart- ments of E. coli was determined for transformants car- rying pKTU1-HCW, pKTU2-HNW or pKTU3-HCT. As judged from the recovered enzymatic activities, the leaderless protein resided largely in the cytoplasm, whereas the plasmids carrying the full-length *aroQ gene directed *MtCM to the periplasmic space (Table 2). To experimentally determine the signal sequence cleavage site of *MtCM in E. coli, the protein variant produced with its native N-terminal signal sequence in Table 1. In vivo complementation of CM deficiency. The selection strain for CM activity, E. coli KA12 ⁄ pKIMP-UAUC [11], was transformed with the plasmids listed, which carry the genes for *MtCM variants or a B. subtilis CM (BsCM; positive control) or no CM (negative control). Growth on minimal medium (M9c+F) in the absence or presence of Tyr (Y) or 0.1 m M salicylate ( ⁄ S) was evaluated after 3 days at 30 °C. Colony sizes were scored on an arbitrary, comparative scale ranging from good (+), moderate (+ ⁄ –), weak (–) to no (0) growth. Plasmid Encoded protein Signal peptide His tag location Growth on minimal medium agar plates M9c+F M9c ⁄ S+F M9c+FY M9c ⁄ S+FY pKTU1-HCW *MtCM Yes C 0 0 + + pKTU2-HNW *MtCM Yes N 0 0 + – a pKTU3-HCT *MtCM No C 0 + + + pMG212H-W BsCM No N + b +++ pMG212H-0 None No None 0 0 + + a High gene expression level appears to be toxic. b Uninduced basal sal promoter activity sufficient for complementation with BsCM. S. Sasso et al. Secreted chorismate mutase from M. tuberculosis FEBS Journal 272 (2005) 375–389 ª 2004 FEBS 377 E. coli KA13 ⁄ pKTU1-HCW was isolated. The pro- tein preparation obtained under denaturing conditions contained the processed and uncleaved forms of *MtCM in a ratio of 10 : 1 as estimated by SDS ⁄ PAGE (not shown). The deconvoluted ESI mass spectrum showed a peak maximum at a M r of 19 536.2. This correlates well with the calculated value for the *MtCM variant lacking the N-terminal 33 resi- dues (predicted, 19 537.4). In parallel, the leaderless *MtCM variant with the additional start methionine as specified by pKTU3-HCT was produced, purified, and subjected to ESI MS. Also in this case, the experi- mental M r correlated well with the theoretical value derived from the sequence (predicted, 19 668.6; observed 19 666). Overproduction and purification of *MtCM variants The leaderless forms of *MtCM with and without a C-terminal His tag were overproduced in a CM-defici- ent E. coli host strain. A specific purification protocol was developed to optimize yield and purity of un- tagged *MtCM by implementing three key features. (a) The highest yields were obtained using strain KA29 (rather than KA13) as the production strain. KA29, which is deficient in thioredoxin reductase, has a more oxidative cytoplasm than wild-type E. coli strains and is therefore recommended for the cytoplasmic produc- tion of proteins with disulfide bonds [30], a probable feature of *MtCM (see Discussion). (b) Heating to 95 °C for 5 min removed the majority of E. coli host proteins (Fig. 2). This step was added because initial thermal denaturation studies showed that most of the *MtCM protein denatured reversibly. (c) Subsequent anion-exchange chromatography under nonbinding conditions (pH 4.5; the calculated isoelectric point of *MtCM is 4.9) eliminated most of the nucleic acid im- purities. In a second anion-exchange chromatography step under binding conditions (pH 8.0), *MtCM was eluted as a sharp peak, affording a highly pure sample (Fig. 2). The final yield of untagged, leaderless *MtCM was 3 mg per liter of bacterial culture (His-tagged *MtCM was obtained at 10 mgÆ L )1 ). The apparent M r values observed for all examined *MtCM species on denaturing polyacrylamide gels run under reducing conditions were generally significantly higher than expected from the sequence. For instance, the 167-residue untagged leaderless *MtCM has a cal- culated M r of 18 603.5 (including the engineered initi- ator Met), in good agreement with the value of 18 602.2 from ESI MS. In contrast, it migrated as a M r 23 000 band on SDS ⁄ PAGE (Fig. 2). Structural characteristics The primary sequence of the translated *aroQ gene was analyzed with the program predictprotein [31], which predicted the secondary structure of the leader- Table 2. Localization of CM activity in subcellular compartments. Variants of *MtCM with and without export leader sequence were pro- duced by CM-deficient E. coli KA13 ⁄ pLysS also carrying one of the plasmids listed. No CM activity was detectable in the absence of an *aroQ gene. CM activities in the isolated fractions of the indicated compartments were determined at 50 l M chorismate and normalized to 1 mL bacterial culture. Plasmid Signal peptide His tag location CM activity [l M converted chorismateÆs )1 Æ(mL culture) )1 ] Cytoplasm Periplasm Medium pKTU1-HCW Yes C 140 (28%) a 330 (66%) 28 (6%) pKTU2-HNW Yes N 270 (18%) a 1100 (75%) 100 (7%) pKTU3-HCT No C 54 (95%) 2.6 (5%) 0.45 (1%) a Residual CM activity may be an artifact from incomplete fractionation of compartments or from partial re-activation during sample prepar- ation of insoluble *MtCM originally present in inclusion bodies. Fig. 2. Purification of untagged, leaderless *MtCM monitored by SDS ⁄ PAGE. Lanes: 1, total cellular protein before IPTG induction of *aroQ expression; 2, after induction; 3, crude lysate (after lysozyme treatment, sonication and removal of insoluble debris); 4, fraction of crude lysate remaining soluble after 5 min at 95 °C; 5, insoluble fraction after heating; M, LMW marker; 6–8, final purified *MtCM, loaded at different concentrations. Secreted chorismate mutase from M. tuberculosis S. Sasso et al. 378 FEBS Journal 272 (2005) 375–389 ª 2004 FEBS less *MtCM to consist of six a-helices, connected by loop segments (Fig. 3A). Such a predominantly a-heli- cal structure was confirmed by CD spectroscopy (Fig. 3B) with the observed troughs at 208.0 and 220.5 nm typical of a-helical proteins [32]. The a-heli- cal content predicted from the primary sequence (67%; Fig. 3B) matches well with the 69% estimated from the CD spectrum [33]. The quaternary structure of *MtCM was investi- gated by gel filtration applying protein samples in a concentration range of 1–61 lm (Fig. 4A). The average apparent M r from five runs was 43 400 ± 3500. Divi- sion by the theoretical subunit M r yields a ratio of 2.3 ± 0.2. The elution volume for *MtCM was independent of the concentration of the injected pro- tein sample (Fig. 4A, inset). Analytical ultracentrifugation of *MtCM was carried out as a complementary experiment to gel fil- tration. Samples with three different *MtCM concen- Fig. 3. Secondary structure of *MtCM. (A) Secondary-structure assignment of the *MtCM sequence [20], using the program Pre- dictProtein [31]. The six a-helices (H1 through H6) predicted for the leaderless *MtCM are indicated (L, loop; E, extended; H, helical structure). Lower case letters denote the signal sequence predicted with the program SIGNALP [27,28]. (B) CD spectrum. The concentra- tion of the leaderless, untagged protein was 0.73 l M in 20 mM potassium phosphate buffer, pH 7.5. h m,r is the mean molar ellipti- city per residue. Fig. 4. Quaternary structure of leaderless, untagged *MtCM. (A) Analytical size-exclusion chromatography. The elution parameter K av of each protein is plotted against the logarithm of the relative molecular mass (M r ) of the standard proteins (s), or of the M r cal- culated from the sequence for a dimeric *MtCM (d). Inset: appar- ent oligomeric state of *MtCM as a function of the concentration of the injected sample. (B) Analytical ultracentrifugation. Shown is the concentration gradient, the computed fit to a dimer model, and the residuals to the fit for the representative sedimentation equilib- rium experiment with 11 l M *MtCM at 19 000 r.p.m. and 20 °C. S. Sasso et al. Secreted chorismate mutase from M. tuberculosis FEBS Journal 272 (2005) 375–389 ª 2004 FEBS 379 trations were run at three different velocities each. A representative sedimentation equilibrium experiment is shown in Fig. 4B. The data fitted in good agreement to a dimer model. The apparent M r values calculated from the nine data sets are summarized in Table 3. The mean value is 35 400 ± 2200, which corresponds to 1.9 ± 0.1 times the calculated M r of the polypep- tide chain. Thermal stability Thermal denaturation and renaturation of *MtCM was followed between 20 and 70 °C by measuring the ellipticity at 222 nm. The denaturation curve shows a sharp transition from the native to the denatured state (Fig. 5), with a calculated melting temperature (T m )of 48 °C. Upon subsequent cooling, the protein renatured easily (Fig. 5). Kinetic studies The CM activity of the leaderless, untagged form of *MtCM was also measured in vitro. The enzyme fol- lows Michaelis–Menten kinetics with the catalytic parameters k cat ¼ 50 s )1 and K m ¼ 180 lm (at 30 °C and pH 7.5). As shown in Table 4, the corresponding values for the C-terminally His-tagged enzyme deviated only slightly. The k cat parameter increases slightly with increasing pH (between pH 5 and pH 8), whereas K m increases more dramatically by a factor of over 20-fold over the same range (Fig. 6A). As a consequence, the catalytic efficiency, k cat ⁄ K m , drops by two orders of magnitude between pH 5 and pH 9. Inclusion of 1 mm tryptophan, phenylalanine or tyro- sine, or 0.5 mm salicylate in the kinetic assays did not alter CM activity by more than 10% (data not shown). Established CM transition state analog inhibitors, which include the oxabicyclic carboxylic acids 1 [34] and 2 [35,36], and adamantane-1-phosphonate 3 [37,38] (Fig. 6B), were tested for their impact on *MtCM activ- ity. Whereas compounds 2 and 3 did not inhibit *MtCM up to concentrations of 100 lm and 1 mm, respectively (data not shown), compound 1 showed competitive inhibition with a K i of 3.7 lm (Fig. 6C). Discussion Our data establish that the ORF Rv1885c from the genome of M. tuberculosis encodes an exported CM. The enzyme was overproduced in E. coli and subjected to detailed structural and functional studies. Sequence Table 3. Analytical ultracentrifugation of *MtCM. Sedimentation equilibrium experiments were carried out at three different protein concentrations at three different velocities each. Listed are the M r values calculated from the nine data sets. Rotor velocity (r.p.m.) Calculated M r at different *MtCM concentrations 1.0 l M 3.0 lM 11 lM 13 000 33 400 32 900 36 900 16 000 34 500 35 400 36 300 19 000 33 000 39 500 36 900 Table 4. Kinetic parameters of *AroQ proteins in comparison with other CMs. Abbreviations: *MtCM-C, leaderless *MtCM with C-ter- minal His tag; *StCM, *PaCM and *EhCM are the *AroQ homologs in S. typhimurium, P. aeruginosa and E. herbicola, respectively. EcCM and BsCM are cytoplasmic CMs described in the text. Assays were performed in 50 m M potassium phosphate buffer, pH 7.5, at 30 °C, unless stated otherwise. k cat was calculated per active site. The inhibition constant K i is listed for transition state analog 1. ND, Not determined. Enzyme k cat (s )1 ) K m (lM) k cat ⁄ K m (M )1 Æs )1 ) K i (lM) *MtCM a 50 ± 2 180 ± 10 2.7 · 10 5 3.7 *MtCM-C a 56 ± 1 150 ± 10 3.7 · 10 5 ND *StCM b 8.9 142 6.3 · 10 4 ND *PaCM b 6.4 98 6.5 · 10 4 ND *EhCM b 9.7 169 5.7 · 10 4 ND EcCM c 64 ± 3 390 ± 51 1.6 · 10 5 2.3 BsCM c 41 ± 2 74 ± 8 5.5 · 10 5 1.0 a The standard deviation indicated was calculated from duplicate measurements. b Measured at 32 °C; data from [12]. c From [36]. Fig. 5. Thermal denaturation of *MtCM. The concentration of the leaderless, untagged protein was 0.73 l M in 20 mM degassed potassium phosphate buffer, pH 7.5. The CD signal was followed at 222 nm for the same sample during heating (s) and subse- quently during cooling (h). h m,r,222 is the mean molar ellipticity per residue at 222 nm. Secreted chorismate mutase from M. tuberculosis S. Sasso et al. 380 FEBS Journal 272 (2005) 375–389 ª 2004 FEBS analysis and CD spectroscopy showed that the mature (leaderless) *MtCM is an a-helical AroQ protein. The polypeptide is predicted to fold into six a-helices, connected by loop segments (Fig. 3A). In contrast, each of the two identical subunits of EcCM adopts only three a-helices which combine to form an inter- twined dimer (Fig. 1B) [8]. The protein sequence encompassing the first three predicted helices of *MtCM aligns well with EcCM (Fig. 7A). However, the first 12 amino acids of EcCM, and thus part of the very long H1-helix which contains the active-site resi- due Arg11 (Fig. 7B), are missing from *MtCM H1. Interestingly, the sequence predicted to form the H4- helix of *MtCM (Fig. 3A) aligns reasonably well with the first part of EcCM H1 (Fig. 7A). This stretch in *MtCM includes a match (Arg134) to Arg11 of EcCM and a pattern of hydrophobic residues which are well conserved among AroQ proteins [4] and which provide helix–helix contacts in EcCM [8]. Indeed, combinato- rial mutagenesis and selection experiments have shown that Arg134 is functionally essential in *MtCM Fig. 7. Conservation and sequence location of presumed active-site residues in *MtCM. (A) Alignment of the sequences of *MtCM (mature form) and EcCM. Residues lining the active site in EcCM and the (presumably) homologous residues in *MtCM are shown in bold. Underlined residues indicate the a-helical regions: for EcCM, this assignment is based on the structurally resolved residues 5–95 in the crystal [8] while for *MtCM, the predicted locations are used (Fig. 3). (B) Scheme of the active site of EcCM, complexed with transition state analog 1 [8]. Fig. 6. Kinetic investigation of leaderless, untagged *MtCM. (A) pH dependence of k cat (d), K m (j) and the catalytic efficiency k cat ⁄ K m (e). The high K m value at pH 8.7 only allowed the determination of k cat ⁄ K m because of the limitation of the maximum chorismate con- centration to 1.3 m M. (B) CM transition state analog inhibitors used in this work: 1 [34]; 2 [35]; 3, adamantane-1-phosphonate [38]. (C) Lineweaver–Burk plot [70] with inhibitor 1. Chorismate concentra- tions were varied in the CM assays at fixed inhibitor concentrations of 0 l M (d), 0.75 lM (j), 1.5 lM (r), 3.0 lM (s), 6.0 lM (h), and 12 l M (e)in50mM potassium phosphate buffer, pH 7.5. Inset: replot of the slopes of the Lineweaver–Burk plot. S. Sasso et al. Secreted chorismate mutase from M. tuberculosis FEBS Journal 272 (2005) 375–389 ª 2004 FEBS 381 (unpublished work), as is its presumed counterpart Arg11 in EcCM [39]. These findings hint at *AroQ proteins being topologically rearranged alternatives of other typical members of the AroQ class. What might be the role of helices H5 and H6? Although the melting temperature of *MtCM is only 48 °C and significantly lower than the 63 °C measured for its mesophilic AroQ homolog EcCM and much lower than the 88 °C for the CM of the thermophile Methanococcus jannaschii [4], we found that *MtCM rapidly renatures after heat denaturation. Preliminary studies with reductants suggest that an intramolecular disulfide bond located in the last two helices (formed between Cys160 and Cys193) contributes to this effi- cient refolding (unpublished work), a feature that was exploited to eliminate most E. coli proteins during *MtCM purification. Thus, the two extra helices might have a stabilizing role. The results of analytical ultra- centrifugation and size-exclusion chromatography experiments show that *MtCM is clearly a dimer. In these experiments, which were carried out over a range of protein concentrations, the ratios of observed M r to calculated subunit M r values were 1.9 (± 0.1) and 2.3 (± 0.2), respectively, and there was no indication of an equilibrium with other quaternary states. The slightly higher value obtained by gel filtration could be explained by an elongated shape of the a-helical dimer; similar observations were previously made for other AroQ proteins [4,40]. To our knowledge, all character- ized members of the AroQ family share a dimeric structure [4,8,9], including the isochorismate pyruvate lyase (PchB) from P. aeruginosa [40]. The permutated helix topology in *AroQ proteins will, however, require a different mode of subunit interaction from that observed in the prototype EcCM. Thus, helices H5 and H6, which are found exclusively in *AroQ subclass members, may also contribute to protein packing in the context of the alternative dimerization interface. Kinetic studies show that *MtCM is a very active CM. Its catalytic parameters are comparable to those of the well-established cytoplasmic enzymes EcCM and BsCM. Moreover, the k cat of *MtCM is fivefold to eightfold higher than k cat of three previously character- ized *AroQ homologs, which results (at similar K m val- ues) in a significantly higher catalytic efficiency (Table 4). As shown in Fig. 7, residues that line the active site in EcCM are in general well conserved in *MtCM. The fact that the K m significantly increases between pH 5 and 9 is consistent with *MtCM having a glutamate (Glu109) at the position corresponding to Gln88 of EcCM (Fig. 7). An analogous pH dependence of K m has been reported for the Gln88Glu variant of EcCM [39,41]. The allosteric CM from S. cerevisiae, the other AroQ protein for which the crystal structure is known, also has a glutamate at the homologous posi- tion, and this residue was shown to be responsible for the strong pH dependence of the CM activity [42]. It is noteworthy that *MtCM is inhibited by transition state analog 1 to the same extent as EcCM and BsCM (Table 4). In contrast, compound 2, which is known to be a 24-fold more selective inhibitor for BsCM than for EcCM [36], did not appreciably inhibit *MtCM. Taken together, these results are consistent with the active site of *MtCM being structurally and function- ally similar to that of a typical AroQ enzyme (Fig. 7B). In vivo assays in E. coli have clearly demonstrated that M. tuberculosis *aroQ can successfully complement the CM deficiency of a heterologous host. Comple- mentation was, however, only observed in constructs devoid of the leader sequence (Table 1). This can be rationalized by the fact that the normal site of action for a CM in E. coli is the cytoplasm, where the enzymes of the shikimate pathway are located. The data from the complementation assays, the cellular fractionation (Table 2), and ESI MS of the exported protein confirm that the predicted signal sequence of *MtCM is recog- nized and cleaved in E. coli. The leader sequence also possesses all of the typical features of signal peptides from M. tuberculosis [43]. The matching criteria include (a) its total length (33 residues), (b) the length of its N-terminal, central hydrophobic, and C-terminal regions (encompassing nine, 15, and nine residues, respectively), (c) the presence of alanines at the ante- penultimate and last positions (AXA motif), and (d) the fact that the mature protein starts with aspartate [43]. From the combined experimental and signal pep- tide prediction data available, we presume that *MtCM is exported by M. tuberculosis, too. As this bacterium is Gram-positive, any transfer of a nonmembrane protein from the cytoplasm must occur directly into the sur- rounding medium rather than into the periplasmic space, the target compartment of the *AroQ homologs from E. herbicola [13], S. typhimurium and P. aeruginosa [12]. The biological significance of an exported CM remains mysterious, however. M. tuberculosis is able to grow in minimal medium lacking aromatic amino acids [44,45]. In its genome, all seven genes corres- ponding to the biosynthetic steps leading from d-erythrose 4-phosphate and phosphoenolpyruvate to chorismate have been identified. Most steps from chor- ismate to the aromatic amino acids were also assigned (see Fig. 1A), including genes for a cytoplasmic pre- phenate dehydratase and a putative cytoplasmic prephenate dehydrogenase [20,46] (http://genolist. pasteur.fr/TubercuList). Although no gene encoding Secreted chorismate mutase from M. tuberculosis S. Sasso et al. 382 FEBS Journal 272 (2005) 375–389 ª 2004 FEBS a cytoplasmic CM in M. tuberculosis has yet been assigned, sequence similarity searches [26,47] have revealed a candidate (Rv0948c) which is currently being examined. Moreover, the secreted *MtCM des- cribed in this work appears unsuitable to close the gap between chorismate made by the cytoplasmic shiki- mate pathway and the terminal (cytoplasmic) branches from prephenate towards Phe and Tyr. Interestingly, Jensen and coworkers found evidence for a mini bio- synthetic pathway from chorismate to phenylalanine in the periplasm of some bacterial species, which utilizes an *AroQ CM and periplasmic versions of prephenate dehydratase (or cyclohexadienyl dehydratase, *PheC) and aromatic amino-acid aminotransferase (*Aat) [12]. However, the absence of *PheC and *Aat homologs from M. tuberculosis argues against a contemporary role for *MtCM in such extracellular Phe biosynthesis, the biological significance of which is also still obscure [12]. It is possible that *MtCM may play some role in the pathogenesis of M. tuberculosis. In both M. tuber- culosis and M. bovis,*aroQ is the second gene in an operon consisting of seven ORFs. While no charac- terized homolog is known for the fourth gene (Rv1883c), the putative intracellular or membrane- bound products of genes 5 (Rv1882c), 6 (Rv1881c) and 7 (Rv1880c) exhibit similarities to a short-chain alcohol dehydrogenase, a membrane lipoprotein (LppE) and a cytochrome P450 (Cyp140), respectively (http://genolist.pasteur.fr/TubercuList). In contrast with these genes, for which information on function in Mycobacterium is still lacking, the genes immediately flanking *aroQ were experimentally investigated and are believed to be disease-related. The first gene in the operon (fbpB, Rv1886c) encodes a secreted mycolyl- transferase (FbpB), which is involved in the final assembly of the mycobacterial cell wall [48], an essen- tial barrier responsible for disease persistence and shielding of the organism from many antibiotics [49]. FbpB is also one of the predominant exported proteins and a constituent of the antigen 85 complex known for its strong fibronectin binding [50,51]. It may play a key role in the invasion of human macrophages, which then leads to intracellular bacterial multiplication or to the notorious dormant infection by M. tuberculosis that affects one third of the world’s population [25]. For these reasons, FbpB is considered an essential tar- get for the development of antimycobacterial chemo- therapeutic agents [48,51]. Immediately downstream of *aroQ is rpfC, which encodes a resuscitation-promo- ting factor that was shown to shorten the lag phase in liquid cultures of closely related mycobacterial species [52]. RpfC has thus been implicated in promoting growth of dormant M. tuberculosis cells [52], which might subsequently lead to open tuberculosis. There are also hints that *AroQ proteins themselves could be pathogenicity factors. For instance, even though animals are believed to lack the shikimate path- way and need to take up essential aromatic compounds from their diet [53], an *aroQ gene encoding a secreted CM was isolated from a cDNA library prepared from the esophageal gland region of the root-knot nematode Meloidogyne javanica [15]. This obligate plant parasite induces the production of giant feeder cells in the root of its host plant presumably by injecting esophageal gland secretions through its stylet. Furthermore, this CM, if produced recombinantly in soybean root cells, suppressed lateral root development and vascular tissue formation in the plant, phenotypes that are also observed during infection [54]. The *aroQ gene of Heterodera glycines, the soybean cyst nematode, was shown to be polymorphic, and this *aroQ polymorph- ism clearly correlated with the virulence of different nematode inbred lines [17]. As the nematode CMs are probably injected into the plant’s cytoplasm [15], a model was presented in which the parasitic enzyme alters the balance of metabolic fluxes within the plant cell to favor establishment of phytonematodic parasit- ism [54]. This hypothesis could also be valid for other phytopathogens found to possess *aroQ genes such as the bacteria E. herbicola [13], Pseudomonas syringae, Ralstonia solanacearum, Xantomonas campestris and Xylella fastidiosa, but not for pathogens of mammals, as their hosts do not possess the shikimate pathway. Apart from Mycobacterium,*aroQ genes are present in several other bacterial pathogens of mammals inclu- ding Burkholderia fungorum, P. aeruginosa, Rhodococ- cus equi, Salmonella typhi, S. typhimurium and Yersinia pestis ([12], unpublished work). Evidence that *AroQ proteins might contribute to virulence in animals was provided by a study with S. typhimurium; the *aroQ promoter was one of four promoters that were found to be induced after infection of mice [18]. In R. equi,a pulmonary pathogen of foals and humans, the *aroQ gene was found to reside in the pathogenicity island on the virulence plasmid [19]. However, the mechanism by which secreted bacterial CMs such as *MtCM may support the pathogenicity process is not known. It will be interesting to examine hypotheses about its involve- ment in the synthesis of compounds that aid in colon- izing the host, for instance, by interaction or interference with coinfecting pathogens, or by engage- ment with the host’s immune system. An understand- ing of the function of *AroQ proteins may ultimately lead to better concepts for fighting diseases such as tuberculosis. S. Sasso et al. Secreted chorismate mutase from M. tuberculosis FEBS Journal 272 (2005) 375–389 ª 2004 FEBS 383 Experimental procedures Materials and general procedures Chromosomal DNA of M. tuberculosis strain H37Rv was obtained from R Brosch (Institut Pasteur, Paris, France). Oligonucleotides were custom-synthesized by Microsynth (Balgach, Switzerland). DNA sequencing was performed on an ABI PRIZM 310 or 3100-Avant Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) by the chain termination method [55], using the BigDye Terminator Cycle Sequencing Kit from the same company. Chorismate and transition state analogs 1 [34], 2 [35] and 3 [38] (Fig. 6B) were prepared by D Ku ¨ nzler [56], R Pulido [57], A Mandal [36] and S Raillard [37], respectively, following published protocols. Cloning techniques and general media compositions were according to standard procedures [58]. For the calculation of M r , molar absorption coefficients and isoelectric points from the primary sequence, the pro- gram package from Genetics Computer Group, Inc. (Madi- son, WI, USA) was used. Strains and plasmid vectors General cloning was carried out in E. coli strain XL1-Blue (Stratagene, La Jolla, CA, USA). The in vivo selection sys- tem KA12 ⁄ pKIMP-UAUC was described previously [11]. E. coli strain KA13 [4,59] was used for *MtCM localization experiments, and strain KA29 for protein overproduction. KA29 carries on its chromosome an isopropyl thio-b-d-gal- actoside (IPTG)-inducible gene coding for T7 RNA polym- erase [60] to allow high-level gene expression controlled by a T7 promoter. It was constructed by site-specific integra- tion of prophage kDE3 into the chromosome of E. coli strain KA25 using the kDE3 Lysogenization Kit from Nov- agen (Madison, WI, USA). KA25 was derived from E. coli strain KA19 by generalized P1 transduction to delete the recA gene employing a JC10289 ⁄ pKY102 lysate [61,62]. The construction of the precursor KA19 will be detailed elsewhere; briefly, it is derived from the thioredoxin reduc- tase-deficient E. coli strain AD494 [30,63] by chromosomal replacement of both endogenous E. coli CM genes tyrA and pheA by a fragment carrying the cat gene of pKIMP- UAUC [11] as well as its tyrA* and pheC genes, which were placed under tac promoter control. The genotype of KA29 is consequently P1 – , D(srlR-recA)306::Tn10 (Tet R ), D(pheA– tyrA)::[tyrA*-pheC-cat (Cam R )], trxB::kan (Kan R ), D(ara- leu)7697, araD139, DlacX74 , galE, galK, rpsL (Str R ), phoR, D(phoA)PvuII, DmalF3, thi, F¢[lac-pro, lacI q ], k (DE3) [(lacUV5-expressed) T7-RNA-pol gene, imm21, Dnin5, Sam7(int-)]. Plasmids pMG212H-W and pMG212H-0, which were used as positive and negative controls, respectively, have been described previously [64]. Plasmid pLysS was pur- chased from Novagen. The high-copy number plasmid pMG211 is a derivative of pKSS [65]. It features the dual- promoter system described previously [64] with a sal and T7 promoter in tandem, the latter being lac repressor- controlled and thus IPTG inducible. The nahR gene on pMG211 codes for an activator of the sal promoter and renders it salicylate inducible. Cloning of a gene into pMG211 using the XhoI site yields a protein with a C-terminal His 6 tag, connected by a Leu-Glu linker. pMG211 also carries a T7 terminator and an ampicillin resistance gene. pMG211 (4690 bp) was constructed in two steps. First, the 2838 bp PshI–BsaI fragment of pMG208 [64] was ligated to the 952 bp NspI(NspI overhang blunt- ended by T4-DNA polymerase treatment)–Bsa I fragment from pKSS [65], yielding pMG210. Secondly, the 2652 bp BstEII–XhoI fragment of pMG210 was ligated to the 2038 bp BstEII–XhoI fragment of pMG209 [64], resulting in pMG211. Construction of *aroQ expression plasmids The M. tuberculosis gene *aroQ was cloned using chromo- somal DNA as the template for PCR. Several different PCR experiments were carried out to obtain plasmids pos- sessing either the entire *aroQ gene with the native signal sequence (affix –W) or a truncated gene without this sequence but having instead an added initiator Met codon (affix –T). The encoded protein either carried a C-terminal His 6 tag linked by Leu-Glu (second affix –C) or an N-ter- minal His 6 tag fused to the presumed leader peptide by a Ser-Ser-Gly linker (affix –N), or no tag (no additional affix). Affix –H denotes a high-copy-number plasmid. The plasmids were assembled as follows: the 600 bp NdeI– XhoI PCR fragment with oligonucleotides 133-TUS (ACC GATGT CATATGCTTACCCGTCCACGTGAGATATA; restriction site used for cloning underlined) and 134-TUN (CGATAAT CTCGAGGGCCGGCGGTAGGGCCTGGC AAT) was ligated to the 4561 bp NdeI–XhoI fragment of pMG211, yielding pKTU1-HCW (5161 bp). The 635 bp NdeI–SpeI PCR fragment with oligonucleotides 135-TUS (ACCGATGT CATATGCACCATCATCATCATCATTCTT CTGGTATGCTTACCCGTCCACGTGAGATATAC) and 136-TUN (CGATAC ACTAGTTATTAGGCCGGCGGTA GGGCCTGGCAAT) was ligated to the 4529 bp NdeI–SpeI fragment of pMG211, yielding pKTU2-HNW (5164 bp). The 504 bp NdeI–XhoI PCR fragment with oligonucleotides 134-TUN and 137-TUS (ACCGATGT CATATGGAC GGCACCAGCCAGTTAGCCGAGTT) was ligated to the 4561 bp NdeI–XhoI fragment of pMG211, yielding pKTU3- HCT (5065 bp). The 509 bp NdeI–SpeI fragment of the PCR with oligonucleotides 136-TUN and 137-TUS was ligated to the 4529 bp NdeI–SpeI fragment of pMG211, yielding pKTU3-HT (5038 bp). All plasmid segments derived from PCR were checked by DNA sequencing using oligonucleotides T7PRO2 (TAATACGACTCACTATA GGG) and 131-TERM (CCCTCAAGACCCGTTTAGA). Secreted chorismate mutase from M. tuberculosis S. Sasso et al. 384 FEBS Journal 272 (2005) 375–389 ª 2004 FEBS [...]... (1997) Thermodynamics of the conversion of chorismate to prephenate: experimental results and theoretical predictions J Phys Chem B 101, 10976– 10982 3 Lee AY, Stewart JD, Clardy J & Ganem B (1995) New insight into the catalytic mechanism of chorismate mutases from structural studies Chem Biol 2, 195–203 4 MacBeath G, Kast P & Hilvert D (1998) A small, thermostable, and monofunctional chorismate mutase from. .. chorismate mutase from the archaeon Methanococcus jannaschii Biochemistry 37, 10062–10073 5 Chook YM, Ke H & Lipscomb WN (1993) Crystal structures of the monofunctional chorismate mutase from Bacillus subtilis and its complex with a transition state analog Proc Natl Acad Sci USA 90, 8600–8603 6 Chook YM, Gray JV, Ke H & Lipscomb WN (1994) The monofunctional chorismate mutase from Bacillus FEBS Journal... 8 9 10 11 12 13 14 15 16 17 18 subtilis: structure determination of chorismate mutase and its complexes with a transition state analog and prephenate, and implications for the mechanism of the enzymatic reaction J Mol Biol 240, 476–500 Helmstaedt K, Heinrich G, Merkl R & Braus GH (2004) Chorismate mutase of Thermus thermophilus is a monofunctional AroH class enzyme inhibited by tyrosine Arch Microbiol... structure of the buried catalytic pocket of Escherichia coli chorismate mutase J Am Chem Soc 117, 3627–3628 Xue Y, Lipscomb WN, Graf R, Schnappauf G & Braus G (1994) The crystal structure of allosteric choris˚ mate mutase at 2.2-A resolution Proc Natl Acad Sci USA 91, 10814–10818 Strater N, Schnappauf G, Braus G & Lipscomb WN ¨ (1997) Mechanisms of catalysis and allosteric regulation of yeast chorismate mutase. .. chorismate mutase from crystal structures Structure 5, 1437–1452 Kast P, Asif-Ullah M, Jiang N & Hilvert D (1996) Exploring the active site of chorismate mutase by combinatorial mutagenesis and selection: the importance of electrostatic catalysis Proc Natl Acad Sci USA 93, 5043–5048 Calhoun DH, Bonner CA, Gu W, Xie G & Jensen RA (2001) The emerging periplasm-localized subclass of AroQ chorismate mutases, exemplified... Characterization of a chorismate mutase from the potato cyst nematode Globodera pallida Mol Plant Pathol 4, 43–50 Bekal S, Niblack TL & Lambert KN (2003) A chorismate mutase from the soybean cyst nematode Heterodera glycines shows polymorphisms that correlate with virulence Mol Plant Microbe Interact 16, 439–446 Bumann D (2002) Examination of Salmonella gene expression in an infected mammalian host using the FEBS... GA (1982) Inhibition of chorismate mutase activity of chorismate mutase- prephenate dehydrogenase from Aerobacter aerogenes Biochemistry 21, 2778–2781 39 Liu DR, Cload ST, Pastor RM & Schultz PG (1996) Analysis of active site residues in Escherichia coli chorismate mutase by site-directed mutagenesis J Am Chem Soc 118, 1789–1790 40 Gaille C, Kast P & Haas D (2002) Salicylate biosynthesis in Pseudomonas... biosynthesis in Pseudomonas aeruginosa: purification and characterization of PchB, a novel bifunctional enzyme displaying isochorismate pyruvate-lyase and chorismate mutase activities J Biol Chem 277, 21768–21775 41 Zhang S, Kongsaeree P, Clardy J, Wilson DB & Ganem B (1996) Site-directed mutagenesis of monofunctional chorismate mutase engineered from the E coli P-protein Bioorg Med Chem 4, 1015–1020 42... generation of protein database search programs Nucleic Acids Res 25, 3389– 3402 Belisle JT, Vissa VD, Sievert T, Takayama K, Brennan PJ & Besra GS (1997) Role of the major antigen of Mycobacterium tuberculosis in cell wall biogenesis Science 276, 1420–1422 Brennan PJ & Nikaido H (1995) The envelope of mycobacteria Annu Rev Biochem 64, 29–63 Peake P, Gooley A & Britton WJ (1993) Mechanism of interaction of the. .. Johnson CR (1985) An inhibitor of chorismate mutase resembling the transition-state conformation J Am Chem Soc 107, 7792–7793 35 Kangas E & Tidor B (2001) Electrostatic complementarity at ligand binding sites: application to chorismate mutase J Phys Chem B 105, 880–888 36 Mandal A & Hilvert D (2003) Charge optimization increases the potency and selectivity of a chorismate mutase inhibitor J Am Chem Soc . Characterization of the secreted chorismate mutase from the pathogen Mycobacterium tuberculosis Severin Sasso, Chandra Ramakrishnan, Marianne. be rationalized by the fact that the normal site of action for a CM in E. coli is the cytoplasm, where the enzymes of the shikimate pathway are located. The data from the complementation assays, the cellular. report the cloning of an M. tuberculosis gene the product of which is homologous to *AroQ CMs. Expression in B A Fig. 1. Chorismate mutase. (A) Biosynthesis of aromatic compounds via the shikimate