Nội dung

Delineation of the roles of FadD22, FadD26 and FadD29 in the biosynthesis of phthiocerol dimycocerosates and related compounds in Mycobacterium tuberculosis Roxane Siméone1,*, Mathieu Léger1,2, Patricia Constant1,2, Wladimir Malaga1,2, Hedia Marrakchi1,2, Mamadou Daffé1,2, Christophe Guilhot1,2 and Christian Chalut1,2 1 CNRS, IPBS (Institut de Pharmacologie et de Biologie Structurale), Toulouse, France 2 Université de Toulouse, UPS, IPBS, France Keywords fatty acyl-AMP ligase; lipid biosynthesis; Mycobacterium tuberculosis; phenolic glycolipids; phthiocerol dimycocerosates Correspondence C. Chalut, Institut de Pharmacologie et de Biologie Structurale, 205 route de Narbonne, 31077 Toulouse Cedex, France Fax: +33 5 61175994 Tel: +33 5 61175473 E-mail: christian.chalut@ipbs.fr *Present address Unité de Pathogénomique Mycobactérienne Intégrée, Institut Pasteur, Paris Cedex, France (Received 8 March 2010, revised 13 April 2010, accepted 16 April 2010) doi:10.1111/j.1742-4658.2010.07688.x Phthiocerol and phthiodiolone dimycocerosates (DIMs) and phenolic glycolipids (PGLs) are complex lipids located at the cell surface of Mycobacterium tuberculosis that play a key role in the pathogenicity of tuberculosis. Most of the genes involved in the biosynthesis of these compounds are clustered on a region of the M. tuberculosis chromosome, the so-called DIM + PGL locus. Among these genes, four ORFs encode FadD proteins, which activate and transfer biosynthetic intermediates onto various polyketide synthases that catalyze the formation of these lipids. In this study, we investigated the roles of FadD22, FadD26 and FadD29 in the biosynthesis of DIMs and related compounds. Biochemical characterization of the lipids produced by a spontaneous Mycobacterium bovis BCG mutant harboring a large deletion within fadD26 revealed that FadD26 is required for the production of DIMs but not of PGLs. Additionally, using allelic exchange recombination, we generated an unmarked M. tuberculosis mutant containing a deletion within fadD29. Biochemical analyses of this strain revealed that, like fadD22, this gene encodes a protein that is specifically involved in the biosynthesis of PGLs, indicating that both FadD22 and FadD29 are responsible for one particular reaction in the PGL biosynthetic pathway. These findings were also supported by in vitro enzymatic studies showing that these enzymes have different properties, FadD22 displaying a p-hydroxybenzoyl-AMP ligase activity, and FadD29 a fatty acylAMP ligase activity. Altogether, these data allowed us to precisely define the functions fulfilled by the various FadD proteins encoded by the DIM + PGL cluster. Introduction Mycobacterium tuberculosis, the etiological agent of tuberculosis, is responsible for 2 million deaths each year. On the basis of accumulated data from numerous studies, the mycobacterial cell envelope appears to play a fundamental role in pathogenicity. This complex structure has a high lipid content and contains a large variety of lipids with unusual structures [1]. Among the surface-exposed lipids, are two structurally related families, phthiocerol dimycocerosates (DIMs) and glycosylated phenolphthiocerol dimycocerosates, also called phenolic glycolipids (PGLs), which have been shown to contribute to the cell envelope permeability Abbreviations DIM, DIM A and DIM B; DIM A, phthiocerol dimycocerosate; DIM B, phthiodiolone dimycocerosate; FAAL, fatty acyl-AMP ligase; Hyg, hygromycin; Km, kanamycin; PGL, phenolic glycolipid; PGL-tb, PGL from Mycobacterium tuberculosis; p-HB, p-hydroxybenzoyl; p-HBA, p-hydroxybenzoic acid; PKS, polyketide synthase. FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS 2715 Phthiocerol dimycocerosates in M. tuberculosis R. Siméone et al. barrier and to virulence [2–5]. DIMs are composed of a mixture of long chain b-diols, esterified by multimethyl-branched fatty acids, the mycocerosic acids [6] (Fig. 1). These compounds are found in a limited group of slow-growing mycobacteria, including M. tuberculosis and Mycobacterium bovis [6]. The chemical structures of PGLs, named PGL-tb in M. tuberculosis, are very similar to those of DIMs, except that the former compounds harbor a phthiocerol chain x-terminated by an aromatic nucleus, the so-called phenolphthiocerol, which, in turn, is glycosylated (Fig. 1). PGLs are produced by most DIM-producing mycobacterial species but are absent from many M. tuberculosis strains [7]. Recently, a third group of molecules related to PGLs, the p-hydroxybenzoic acid derivatives, has been identified in M. tuberculosis and M. bovis BCG [7]. These compounds, which contain the same glycosylated phenolic moiety as PGLs, are released into the culture medium during in vitro growth. Most of the genes required for DIM and PGL biosynthesis and translocation are clustered in a 70 kb region of the M. tuberculosis chromosome, the DIM + PGL locus [5,8]. Five of these genes (ppsA–E) encode type I polyketide synthases (PKS), which are responsible for the elongation of either C22–24 fatty acids or p-hydroxyphenylalkanoic acids by the addition of malonyl-CoA and methylmalonyl-CoA extender units to yield phthiocerol and phenolphthiocerol derivatives, respectively [9] (Fig. 2). The protein encoded by pks15 ⁄ 1, another type I PKS, catalyzes the elongation of p-hydroxybenzoic acid (p-HBA) with CH 2 CH H3C (CH 2 ) m1 CH C HC HC CH3 CH2 p-1 C CH3 HC DIM A HC p'-1 Me HO O CH 2 CH 2 DIM B p'-1 CH 2 ( CH 2 ) n' CH3 C OMe HC O CH 2 (CH 2 ) m2 OMe Mycoside B (CH 2 )4 CH HC CH3 ( CH 2 ) n O CH3 CH2 CH3 CH2 HC p-1 CH CH3 OMe C O CH2 HC O CH O O CH3 2716 CH3 Common lipid core OH OH O CH3 PGL-tb CH Me HO R (CH 2 ) m2 OMe OMe OMe p-1 CH3 O O CH3 HC ( CH 2 ) n O O HC C CH3 O CH2 CH2 O (CH 2 )4 CH C O CH3 CH2 CH3 Me HO CH O O ( CH 2 ) n' CH3 CH 2 CH H3C (CH 2 ) m1 CH 2 ( CH 2 ) n Me R O CH3 CH2 HC CH CH3 OMe C O CH3 CH2 HC (CH 2 )4 CH O O malonyl-CoA units to form p-hydroxyphenylalkanoic acid derivatives, which, in turn, are used by PpsA–E to yield phenolphthiocerol and its relatives [7]. The lack of PGL-tb in Euro-American isolates of the M. tuberculosis strains results from a natural frameshift mutation within pks15 ⁄ 1 [7,10]. Finally, the protein encoded by mas, the last pks gene of the DIM + PGL locus, catalyzes the iterative elongation of C18–20 fatty acids with methylmalonyl-CoA units to generate mycocerosic acids [11] (Fig. 2). In addition to these pks genes, the DIM + PGL locus contains four genes (fadD22, fadD26, fadD28, and fadD29) encoding FadD proteins that are conserved in all sequenced mycobacterial species producing DIMs and PGLs [8,12]. FadD26, FadD28 and FadD29 belong to a group of long-chain fatty acyl-AMP ligases (FAALs) produced by M. tuberculosis that convert long-chain fatty acids to acyl-adenylates [13]. FadD28 is involved in the formation of mycocerosic acids by activating the Mas substrates [14], whereas FadD26 was shown to be directly involved in DIM biosynthesis by catalyzing the loading of long-chain fatty acids onto PpsA [5,15] (Fig. 2). Recently, Ferreras et al. [16] demonstrated that FadD22 is specific for the formation of PGL, also called mycoside B, in M. bovis BCG. This protein, which harbors an adenylation domain and a C-terminal aroyl carrier protein domain, catalyzes the formation of p-hydroxybenzoylAMP from p-HBA and subsequent transfer of the p-hydroxybenzoyl (p-HB) moiety onto the aroyl carrier protein domain. CH3 CH 2 ( CH 2 ) n' CH3 p'-1 R Fig. 1. Structures of the DIM A and DIM B produced by Mycobacterium tuberculosis and Mycobacterium bovis BCG and of the glycosylated phenolphthiocerol dimycocerosates produced by M. tuberculosis (PGL-tb) and M. bovis BCG (mycoside B). p, p¢ = 3– 5; n, n¢ = 16–18; m1 = 20–22; m2 = 15–17; R = C2H5 or CH3. FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS R. Siméone et al. Phthiocerol dimycocerosates in M. tuberculosis PpsA FadD26 OH O Fatty acid 18–20 ATP PpsB-E ACP KS AT KR ACP R 18–20 Rv2953, TesA AMP + PPi S OH OH S O OCH3 Phthiocerol O OH 18–20 Malonyl 18–20 CO2 AMP + PPi 1) Hydrolysis PKS15/1 FadD22 PpsA 2) FadD29 ATP S ATP S O 13–15 HO HO Malonyl Pyruvate HO AMP + PPi CO2 FadD28 Mas KS AT DH ER KR ACP COOH O 3–5 methylmalonyl OH O – OOC O CH2 Chorismate PapA5 S ATP HO OCH3 13–15 p–hydroxyphenylalkanoate Rv2949c OH OH OH 13–15 COOH p-HBA 13–15 Phenolphthiocerol O 8–9 malonyl 8–9 CO2 R HO Rv2953, TesA AMP + PPi S O HO PpsB-E ACP KS AT KR ACP KS AT DH ER KR ACP 3–5 CO2 3–5 15–17 PGL-tb O Common lipid core 14–16 R Fatty acid 13–15 18–20 O O O DIM 3–5 15–17 Mycocerosic acids OCH3 O 3–5 15–17 Fig. 2. Schematic representation of the roles of the FadD proteins encoded by the DIM + PGL locus in the biosynthesis of DIMs and related compounds in Mycobacterium tuberculosis. The role of the activation enzymes FadD22, FadD26 and FadD29 was determined during this study. R = C2H5 or CH3. KS, ketoacylsynthase; AT, acyltransferase; DH, dehydratase; ER, enoylreductase; KR, ketoreductase; ACP, acyl carrier protein. Despite this progress in our knowledge of the biosynthesis of DIMs and PGLs, the in vivo functions of these FadD proteins in this pathway remain to be established and ⁄ or completed. For instance, the involvement of FadD26 in the biosynthesis of phenolphthiocerol, and thereby in PGL production, has never been investigated, and the biological function of FadD29 is still unknown. In addition, the possible redundancy of these proteins in the activation of the various intermediates in DIM and PGL biosynthesis has not been analyzed. This study was undertaken to dissect the roles of FadD22, FadD26 and FadD29 in the biosynthesis of DIMs and related compounds. Biochemical analyses of an M. bovis BCG mutant harboring a large deletion within fadD26 established that this gene is specific for DIM biosynthesis. By constructing a knockout mutant in M. tuberculosis, we also provide evidence that FadD29 is required for PGL production, and that FadD22 and FadD29 are functionally nonredundant. Consistently, in vitro enzymatic assays revealed that these two proteins have distinct substrate specificities. Results and Discussion FadD26 is specifically involved in DIM biosynthesis A spontaneous M. bovis BCG mutant strain, named PMM137, harboring a 1450 bp deletion (nucleotides 122–1571) within fadD26, was isolated in the laboratory (Fig. 3A). It has been previously shown that disruption of fadD26 in M. tuberculosis strains MT103 and Erdman abolished the production of DIMs [3,5], but the requirement for FadD26 in the formation of PGL-tb was not investigated, because the strains FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS 2717 R. Siméone et al. M 13 7 M 13 7: PM PM BC G M 13 W T pM 26 M 26 D 7 13 M PM 1450 bp PM BC G 26G 26F PMM137 7: p B W T A D Phthiocerol dimycocerosates in M. tuberculosis fadD26 121 bp 181 bp kb 10.0 6.0 4.0 3.0 2.0 1.5 1.0 PM M 1 BC 37 G W T DIM A DIM B PGL (mycoside B) 20 10 0 1280 1360 1440 1520 40 30 20 10 1600 Mass (m/z) 0 1280 1502.62 1418.52 1432.54 1446.55 1460.57 1474.59 1390.49 1348.44 1362.46 1376.47 50 1360 1440 1516.64 30 60 BCG WT DIM A 1488.60 40 70 1404.51 50 80 1306.39 1320.40 1334.42 60 90 1488.47 1390.36 1404.37 1418.39 70 1348.31 1362.33 1376.34 Intensity (%) 80 100 Intensity (%) 90 PMM137:pM26D DIM A 1502.48 100 1516.52 C 1432.40 1446.42 1460.44 1474.45 0.5 1520 1600 Mass (m/z) Fig. 3. Genetic and biochemical characterization of the Mycobacterium bovis BCG fadD26 mutant strain. (A) Schematic diagram of the genomic organization of the fadD26 locus in the M. bovis BCG fadD26 (PMM137) mutant strain. Black boxes represent portions of the fadD26 gene that are still present in the PMM137 chromosome, and the hatched box represents the portion of the fadD26 gene that was deleted. The fadD26 gene from PMM137 and M. bovis BCG wild-type (WT) were PCR-amplified using primers fadD26F (26F) and fadD26G (26G). The PCR products were separated on an agarose gel and analyzed by sequencing. The resulting sequences revealed a 1450 bp deletion (nucleotides 122–1571) within the PMM137 fadD26 gene. (B) TLC analyses of radiolabeled DIMs (left panel) and glycoconjugates (right panel) extracted from wild-type M. bovis BCG, from PMM137, and from PMM137 complemented with pM26D. For DIM analysis, lipid extracts were loaded onto a TLC plate run in petroleum ether ⁄ diethylether (90 : 10, v ⁄ v), and visualized by using a PhophorImager system. For glycoconjugate analysis, lipids were loaded onto a TLC plate run in CHCl3 ⁄ CH3OH (95 : 5, v ⁄ v), and visualized by spraying the TLC plate with 0.2% (w ⁄ v) anthrone in concentrated H2SO4 followed by heating. The positions of DIM A, DIM B and PGL (mycoside B) are indicated. (C) MALDI-TOF mass spectra of the purified lipid exhibiting similar TLC mobility as DIM A from M. bovis BCG PMM137:pM26D (left panel) and from wild-type M. bovis BCG (right panel). A similar analysis was performed with the lipids exhibiting the mobility of DIM B in PMM137:pM26D. A MALDI-TOF spectrum similar to that of DIM B from wild-type M. bovis BCG was obtained (data not shown). used in these studies were naturally deficient in PGL production. To determine the role of fadD26 in the biosynthesis of DIMs and PGL in M. bovis BCG, lipids were extracted from PMM137 and analyzed by TLC. Disruption of fadD26 in M. bovis BCG abolished the production of phthiocerol dimycocerosate (DIM A) 2718 and of phthiodiolone dimycocerosate (DIM B), a DIM structural variant that contains a keto group in place of the methoxy group at the terminus of the b-diols [17] (Figs 1 and 3B, left panel). In sharp contrast, PMM137 was still able to produce, although at a lower level than in the wild-type strain, a major glycoconjugate exhibiting identical TLC mobility to that of FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS R. Siméone et al. mycoside B, the species-specific PGL of M. bovis (Figs 1 and 3B, right panel). The nature of this product was confirmed by MALDI-TOF MS analysis (Fig. S1). This finding suggested that fadD26 might be specific for the production of DIMs in the various DIM-producing species of mycobacteria. To establish the relationship between mutation of fadD26 and the absence of DIMs in PMM137, complementation studies were performed by transferring a plasmid, named pM26D, containing a wild-type copy of fadD26, into the mutant strain. Complementation resulted in the production of new products exhibiting TLC mobilities similar to those of DIM A and DIM B produced by the wild-type strain (Fig. 3B). The nature of these compounds was established by MALDITOF MS analysis, which confirmed that the new products accumulated by the complemented strain corresponded to DIMs (Fig. 3C). We also noticed that the production of DIMs in the PMM137:pM26D strain was lower than that in the wild-type strain (Fig. 3B), suggesting partial restoration of DIM biosynthesis in the complemented strain. This partial restoration could be explained by weak expression of fadD26 from pM26D and ⁄ or by a partial polar effect of the fadD26 deletion on the expression of the downstream ppsA–E genes, which belong to the same operon [5]. The latter hypothesis was supported by the observation that PMM137 and the PMM137:pM26D strain produced significantly less PGL than did the wild-type strain (Fig. 3B, right panel). The ppsA–E genes are indeed required for DIM and PGL biosynthesis [9], and the presence of lower amounts of PpsA–E in the bacterial cell would decrease the production rates of both classes of lipids. In such a case, introduction of a wild-type allele of fadD26 into PMM137 would complement the disruption of fadD26 but might not counteract the polar effect on the ppsA–E genes, leading to the production of lower amounts of DIMs and PGL in the complemented strain than in the wild-type strain. In agreement with this hypothesis, a similar polar effect on the expression of the downstream ppsA–E genes was observed in a fadD26 M. tuberculosis mutant [5]. FadD26 was previously shown to catalyze the formation of acyl-AMP from long-chain fatty acids and their subsequent transfer to PpsA in vitro [13,15]. Our data further demonstrate that, in vivo, FadD26 specifically activates C22–24 fatty acyl chains that are loaded onto PpsA for the formation of the phthiocerol chain but is not required for the production of PGL in M. bovis BCG. The conservation of the DIM and PGL biosynthetic pathways in the various DIM-producing mycobacteria suggests that FadD26 has similar Phthiocerol dimycocerosates in M. tuberculosis functions in all DIM-producing species, including M. tuberculosis. FadD29 is specifically involved in PGL biosynthesis During the biosynthesis of phenolphthiocerol, both p-HBA and its elongation product, p-hydroxyphenylalkanoic acid, need to be activated by one or several FadD proteins prior to their subsequent transfer onto PKS15 ⁄ 1 and PpsA, respectively [7,9] (Fig. 2). The finding that FadD26 catalyzes the activation and the loading of biosynthetic intermediates onto PpsA during DIM biosynthesis prompted us to examine whether another FadD protein is specifically involved in the activation of the PpsA substrates during the formation of PGLs. Among the fadD genes that belong to the DIM + PGL cluster, we identified a single candidate gene, fadD29, that may perform this function. Indeed, FadD28 was shown to be involved in the formation of mycocerosic acids by activating the Mas substrates [14], and FadD22 was proposed to promote the transfer of the p-HB moiety onto PKS15 ⁄ 1 [16,18]. To examine the involvement of FadD29 in the biosynthesis of DIMs and PGLs, we took advantage of the existence in the laboratory of an M. tuberculosis H37Rv recombinant strain, harboring an unmarked mutation [19] within fadD29 (Fig. S2). This strain, named PMM66 (fadD29::res), was constructed by replacing the wild-type allele of fadD29 with a kanamycin (Km)-disrupted allele using the ts ⁄ sacB procedure and subsequent excision of the res–km–res cassette by site-specific recombination between the two res sites [19,20] (Fig. S2). As M. tuberculosis H37Rv and its derivatives are naturally devoid of PGL-tb, because of a frameshift mutation in pks15 ⁄ 1, this strain was transformed with plasmid pPET1 carrying a functional M. bovis BCG pks15 ⁄ 1 gene [7], and the resulting PMM66:pPET1 strain was labeled with [1-14C]propionate, a precursor known to be incorporated into methyl-branched fatty acyl-containing lipids, including DIMs. Lipids were then extracted from the recombinant strain and analyzed by TLC. These analyses revealed that the strain was still able to produce DIM A and DIM B, the two structural variants of DIMs (Figs 1 and 4A). This indicated that fadD29 is not required for the production of DIMs in M. tuberculosis. We next focused on the putative role of fadD29 in PGL-tb biosynthesis. TLC analysis of lipids from the PMM66:pPET1 strain showed that disruption of fadD29 abolished the production of PGL-tb (Fig. 4B), FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS 2719 B M PM 37 H H 37 R R v: v PM :pP ET M 1 66 :p PE T1 A R. Siméone et al. pP ET 1 66 PM :pP ET M 6 1 PM 6:p PE M T1 66 : PM :pP pR ET S0 M 4 1: 66 pR :p S2 PE 2 T1 :p R S2 3 Phthiocerol dimycocerosates in M. tuberculosis DIM A PGL-tb DIM B 100 20 10 0 1756.6 1866.2 1975.8 30 10 Mass (m/z) 0 1750 1816 1882 1976.48 1948 1990.50 40 1962.46 50 20 2085.4 1934.42 1948.44 1906.38 60 H37Rv:pPET1 PGL-tb 2004.53 2018.54 30 70 1850.33 1864.33 1878.35 1892.37 40 1822.69 1836.70 1850.71 1864.73 1878.75 1892.76 60 Intensity (%) 1934.81 1948.83 70 50 90 80 1962.85 1976.87 1990.88 2004.88 2018.91 80 Intensity (%) PMM66:pPET1:pRS23 PGL-tb 1906.77 90 1920.40 100 1920.80 C 2014 2080 Mass (m/z) Fig. 4. Analyses of lipids extracted from the Mycobacterium tuberculosis H37Rv fadD29::res (PMM66) mutant strain. (A) TLC analysis of radiolabeled DIMs from wild-type M. tuberculosis and from the PMM66 recombinant strain complemented with pPET1. Lipid extracts dissolved in CHCl3 were loaded onto the TLC plate, which was run in petroleum ether ⁄ diethylether (90 : 10, v ⁄ v); lipids were visualized using a PhophorImager system. The positions of DIM A and DIM B are indicated. (B) TLC analysis of glycolipids extracted from wild-type M. tuberculosis complemented with pPET1, from the PMM66:pPET1 recombinant strain and from the PMM66:pPET1 complemented strain. Lipids were dissolved in CHCl3, and the plate was run in CHCl3 ⁄ CH3OH (95 : 5, v ⁄ v). Glycoconjugates were visualized by spraying the TLC plate with 0.2% (w ⁄ v) anthrone in concentrated H2SO4, followed by heating. The position of PGL-tb is indicated. (C) MALDI-TOF mass spectra of purified lipids exhibiting similar TLC mobility as PGL-tb from M. tuberculosis PMM66:pPET1:pRS23 (left panel) and of PGL-tb from wild-type M. tuberculosis (right panel). strongly suggesting that this gene encodes a protein involved in the biosynthesis of PGL-tb in M. tuberculosis strains that contain this glycolipid. Nevertheless, complementation of the fadD29 mutation by transferring a plasmid, pRS04, harboring an intact copy of fadD29, into PMM66:pPET1 did not restore the production of PGL-tb (Fig. 4B). A close examination of the DIM + PGL locus organization revealed that fadD29 is located upstream of Rv2949c, a gene required for the biosynthesis of p-HBA, a precursor of PGL-tb [7,21]. We thus speculated that disruption of fadD29 may exert a polar effect on the expression 2720 of Rv2949c, abolishing the production of p-HBA, which, in turn, would suppress the synthesis of PGL-tb in the mutant strain. To examine whether the absence of PGL-tb in the cell envelope of the PMM66:pPET1 mutant solely relied on a polar effect on the expression of Rv2949c or resulted from both a polar effect on the expression of Rv2949c and disruption of fadD29, the mutant strain was transformed with either a plasmid (pRS22) containing a copy of Rv2949c or a plasmid (pRS23) carrying both the wildtype allele of Rv2949c and that of fadD29. The transfer of Rv2949c alone into the PMM66:pPET1 mutant FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS R. Siméone et al. was not sufficient to restore the production of PGL-tb (Fig. 4B). In contrast, complementation of the PMM66:pPET1 strain with both Rv2949c and fadD29 led to the production of a major glycoconjugate exhibiting identical TLC mobility to that of PGL-tb (Fig. 4B). The MALDI-TOF mass spectrum of this lipid showed a series of pseudomolecular ion (M + Na+) peaks with m ⁄ z values identical to those observed in the mass spectrum of PGL-tb purified from the wild-type strain (Fig. 4C). Therefore, it could be concluded from these studies that the disruption of fadD29 had a polar effect on the expression of Rv2949c in PMM66 and that fadD29 encodes a protein involved in the biosynthesis of PGL-tb in M. tuberculosis. These data also established that, in contrast to fadD26, fadD29 does not play any role in the biosynthesis of DIMs in M. tuberculosis. FadD22 and FadD29 catalyze independent catalytic events in the PGL pathway Our data established that fadD29 is specifically required for the formation of PGL-tb in M. tuberculosis. Interestingly, it has recently been shown that disruption of fadD22 in M. bovis BCG abolished the production of its species-specific PGL, the so-called mycoside B [16]. By analyzing an M. tuberculosis H37Rv recombinant strain harboring a deletion in fadD22, we confirmed the role of fadD22 in the biosynthesis of PGL-tb in M. tuberculosis (data not shown). These data suggest that fadD22 and fadD29 encode nonredundant enzymes that are responsible for one particular reaction in the PGL biosynthetic pathway. To further demonstrate that FadD22 and FadD29 recognize different substrates and therefore activate distinct intermediates (i.e. p-HBA and p-hydroxyphenylalkanoic acid) in vivo, we compared their ability to convert p-HBA and fatty acid substrates into AMP derivatives in vitro. FadD22 has been shown to be able to convert p-HBA into p-HB-AMP and FadD29 has been classified as a FAAL protein [13,16], but these two enzymes were not assayed with different substrates in parallel experiments. Lauric (C12) fatty acid was used as a surrogate substrate in these assays, because the natural substrate, p-hydroxyphenylalkanoic acid, was not commercially available. p-Hydroxyphenylalkanoic acid consists of a C17–19 fatty acid chain attached to a p-hydroxyphenyl moiety, and the enzyme responsible for the activation of this intermediate in vivo is expected to display FAAL activity in vitro. FadD22 and FadD29 were overexpressed in Escherichia coli and purified, and their enzymatic activities were monitored using either radiolabeled lauric (C12) Phthiocerol dimycocerosates in M. tuberculosis fatty acid or p-HBA in the presence of ATP. Incubation of FadD29 in the presence of [14C]lauric acid led to the production of a radiolabeled compound exhibiting an Rf value identical to that of a chemically synthesized C12 fatty acyl-AMP (Fig. 5, left panel). No radiolabeled product could be detected when the reaction was performed with FadD22, indicating that this protein was unable to generate a fatty acyl adenylate from a fatty acid and ATP (Fig. 5, left panel), in contrast to FadD29. As expected, in the presence of [14C]p-HBA, FadD22 catalyzed the formation of a radiolabeled compound that exhibited migration similar to that observed by Ferreras et al. [16] for p-HBAMP (Fig. 5, right panel). These data strongly suggested that, in our assay, FadD22 catalyzed the formation of [14C]p-HB-AMP from [14C]p-HBA. Under these experimental conditions, FadD29 was unable to synthesize a radiolabeled p-HB-AMP compound (Fig. 5, right panel). In both cases, the formation of AMP derivatives (fatty acyl-AMP or p-HBAMP) was dependent on the presence of the enzyme (FadD29 or FadD22) (Fig. 5) and ATP (data not shown), confirming that the detected compounds were the products of enzymatic reactions between the [14C]lauric acid [14C]p-HBA FadD22 FadD29 [14C]p-HBA [14C]lauric acid [14C]C12-AMP [14C]p-HB-AMP Fig. 5. In vitro enzymatic activities of FadD22 and FadD29. The enzymatic activities of FadD22 and FadD29 were determined by incubating each enzyme with either [14C]lauric acid or [14C]p-HBA in the presence of ATP, and monitoring C12-AMP (left panel) or p-HBAMP (right panel) formation using radio-TLC. Radiolabeled C12-AMP products were resolved by running the TLC plate in butan-1-ol ⁄ acetic acid ⁄ water (80 : 25 : 40) at room temperature, and p-HBAMP formation was analyzed with silica gel TLC plates (G60) developed in ethyl acetate ⁄ isopropyl alcohol ⁄ acetic acid ⁄ water (70 : 20 : 25 : 40, v ⁄ v). TLC plates were exposed to phosphor imaging plates and quantificated by use of a PhosphorImager. FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS 2721 Phthiocerol dimycocerosates in M. tuberculosis R. Siméone et al. radiolabeled substance used in the assay and ATP. We therefore concluded that FadD22 and FadD29 exhibit different enzymatic activities, FadD22 being a p-HBAMP ligase and FadD29 a FAAL protein, as suggested in earlier studies [13,16]. These in vitro enzymatic experiments, combined with our genetic and biochemical studies, revealed that FadD22 and FadD29 activate distinct intermediates during the formation of the phenolphthiocerol chain: FadD22 catalyzes the activation of p-HBA and its subsequent transfer onto PKS15 ⁄ 1, to yield p-hydroxyphenylalkanoate; and this latter lipid is then activated by FadD29 and transferred onto PpsA (Fig. 2). Conclusions The enzymatic activities of FadD22, FadD26 and FadD29 had been investigated in previous studies [13,16], but their precise role in vivo was not clearly defined. In this article, we provide novel experimental data that allow us to propose a specific role for each of these enzymes in the PGL and DIM biosynthetic pathways (Fig. 2). Our results clearly demonstrate that FadD22 and FadD29 are functionally nonredundant and activate distinct intermediates during the formation of PGLs, a point that has never been reported before. Indeed, the two enzymes have been found to be independently required for the biosynthesis of PGLs, and our enzymatic studies showed that FadD22 and FadD29 recognize different substrates. In contrast, we established that FadD26 is required for the production of DIMs but not for that of PGLs. This is the first demonstration that, although they have identical enzymatic properties in vitro, i.e. the formation of acyl-adenylates and transfer of their products to a common PKS, namely PpsA, FadD26 and FadD29 play distinct roles in vivo. One possible explanation may be that the enzymatic properties of these enzymes have been determined using fatty acid chains ranging from three to 18 carbons [13,15], whereas, in vivo, the enzymes may exhibit a narrower substrate specificity for the activation of their natural substrates, a C22–24 fatty acid chain for FadD26, and a C17–19 fatty acid chain terminated by a p-hydroxyphenyl moiety for FadD29. Alternatively, substrate specificity might be controlled in vivo by specific interactions between FadD29 and PKS15 ⁄ 1 on the one hand, and FadD26 and the FasI system on the other hand. Further investigations, including protein interaction studies and structural studies, will be necessary to elucidate the precise molecular mechanisms underlying the biosynthesis of DIMs and PGLs. 2722 Experimental procedures Bacterial strains, growth media, and culture conditions Plasmids were propagated at 37 C in E. coli DH5a or E. coli HB101 in LB broth or LB agar (Invitrogen, Cergy Pontoise, France) supplemented with either Km (40 lgÆmL)1) or hygromycin (Hyg) (200 lgÆmL)1). The M. tuberculosis H37Rv and M. bovis BCG wild-type strains and their derivatives were grown at 37 C in Middlebrook 7H9 broth (Invitrogen) containing 0.2% dextrose, 0.5% BSA fraction V, and 0.0003% beef catalase and 0.05% Tween-80 when necessary, and on solid Middlebrook 7H11 broth containing 0.2% dextrose, 0.5% BSA fraction V, and 0.0003% beef catalase and 0.005% oleic acid. For biochemical analyses, mycobacterial strains were grown as surface pellicles on Sauton’s medium. When required, Km and Hyg were used at concentrations of 40 and 50 lgÆmL)1, respectively. Construction of a fadD29 M. tuberculosis H37Rv unmarked mutant An M. tuberculosis H37Rv mutant containing a disrupted fadD29::res-km-res gene (PMM34) on the chromosome was constructed by allelic exchange using the ts ⁄ sacB procedure [20] (Fig. S2). Two DNA fragments overlapping fadD29 were amplified by PCR from M. tuberculosis H37Rv chromosomal DNA, using oligonucleotides fadD29A and fadD29B (Table 1). These PCR fragments were cloned, after insertion of a Km resistance cassette flanked by two res sites from transposon cd [19] between the BamHI and Table 1. Oligonucleotides used in this study. Gene Primer Oligonucleotide sequence (5¢–3¢) fadD22 fadD22C TCACGGGTCGCATCAAGGAGC fadD22J ACAACATATGCGGAATGGGAATCTAGC fadD22K ACAAAAGCTTCTTCCCAAGTTCGGAATCGA fadD26F CATAGTGAACGCCAGAAAGCCG fadD26G TAGGTAGTCGATTAGCCAGTGG fadD26K ACAACATATGCCGGTGACCGACCGTT fadD26L ACAAAAGCTTCATACGGCTACGTCCAGCC fadD29A GCTCTAGAGTTTAAACCGCGCTCGGGGTACCTGG fadD29B GCGCGGCCGCGTTTAAACCGATCGCGCAGCGCATC fadD29C TCGCGACGACGTGGAAGAGGC fadD29D ATCGGTTCGTAGCCTCCAGGC fadD29E CCGACTCGGATTCGTATGAAAG fadD29F GTTATGCCATAGCATCTAGGC fadD29I ACTTCGCAATGAAAACCAACTCGTCATTTC 2949H ACTTCGCAATGACCGAGTGTTTTCTATCTG 2949I ACAAAAGCTTTATTGGATGACCGCCCTAG res1 GCTCTAGAGCAACCGTCCGAAATATTATAAA res2 GCTCTAGATCTCATAAAAATGTATCCTAAATCAAATATC fadD26 fadD29 Rv2949c res FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS R. Siméone et al. Phthiocerol dimycocerosates in M. tuberculosis EcoRV restriction sites, into the mycobacterial thermosensitive suicide plasmid pPR27 [20]. The resulting plasmid was transferred by electrotransformation into M. tuberculosis, and allelic exchange at the fadD29 locus was screened by PCR analysis of the genomic DNA from several Km-resistant and sucrose-resistant colonies by using a set of specific primers (fadD29C, fadD29D, fadD22C, res1, and res2; Table 1) (Fig. S2). To recover the Km resistance cassette from M. tuberculosis PMM34, this strain was transformed with the thermosensitive plasmid pWM19, which contains the resolvase gene of transposon cd and a Hyg resistance gene [19]. Transformants were resuspended in 5 mL of Middlebrook 7H9 broth, and incubated for 48 h at 32 C to allow the expression of Hyg resistance. Hyg was then added to the transformation mixtures, and cells were incubated at 32 C for 12 days. Serial dilutions of bacterial cultures were then plated onto Middlebrook 7H11 plates and incubated further at 39 C. Several colonies were picked and tested for growth on plates containing Km. Several clones that were unable to grow on Km-containing plates but that showed normal growth on control antibiotic-free plates were selected and analyzed by PCR, using primers fadD29C and fadD29D (Fig. S2). One clone giving the corresponding pattern for excision of the Km resistance cassette in fadD29 was selected for further analyses, and named PMM66 (fadD29::res) (Table 2). Construction of complementation and expression plasmids For the construction of pRS04, a region covering fadD29 plus 70 bp of the sequence upstream of the start codon was PCR-amplified from M. tuberculosis H37Rv genomic DNA, using oligonucleotides fadD29E and fadD29F (Table 1). The PCR product was purified and inserted into the XmnI site of pMV361 [22]. To construct pRS22 and pRS23, regions covering Rv2949c (for pRS22) or fadD29 plus Rv2949c (for pRS23) were PCR-amplified from M. tuberculosis H37Rv genomic DNA, using oligonucleotides 2949H and 2949I (for pRS22) or fadD29I and 2949I (for pRS23). PCR products were digested with HindIII, and cloned between the XmnI and HindIII sites of pMV361 downstream of the phsp60 promoter. The complementation vector pM26D was constructed by amplifying fadD26 from M. bovis BCG genomic DNA, using oligonucleotides fadD26K and fadD26L (Table 1). The PCR product was digested with NdeI and HindIII, and cloned between the NdeI and HindIII sites of pMV361e, a pMV361 derivative containing the pblaF* promoter instead of the original phsp60 promoter and carrying a Km resistance marker. For production of recombinant M. tuberculosis FadD22 in E. coli, fadD22 was amplified by PCR from M. tuberculosis H37Rv total DNA, using oligonucleotides fadD22J and fadD22K (Table 1), and the resulting fragment was inserted into the pET26b E. coli expression vector (Novagen, Madison, WI, USA) under control of the T7 promoter, to give plasmid pETD22. To produce recombinant M. tuberculosis FadD29 in E. coli, we used pRT11, a derivative of pET21a containing fadD29 of M. tuberculosis (a generous gift from R. S. Gokhale, National Institute of Immunology, New Delhi, India). These vectors allow the production of recombinant FadD22 and FadD29 fused to a poly-His tag at their C-termini in E. coli BL21. Extraction and purification of DIMs and PGLs For each strain, bacterial cells were harvested from 200 mL cultures, and sterilized by adding CHCl3 ⁄ CH3OH (1 : 2, v ⁄ v) for 2 days at room temperature. Lipids were extracted twice with CHCl3 ⁄ CH3OH (2 : 1, v ⁄ v), washed twice with water, and dried before analysis. Extracted mycobacterial lipids were analyzed by TLC after resuspension in CHCl3 at a final concentration of 20 mgÆmL)1. Equivalent amounts of lipids from each extraction were spotted onto silica gel G60 plates (20 · 20 cm; Merck, Darmstadt, Germany), and separated with either petroleum ether ⁄ diethylether (90 : 10, v ⁄ v) for DIM analysis or CHCl3 ⁄ CH3OH (95 : 5, v ⁄ v) for PGL analysis. DIMs and PGLs were Table 2. Strains and plasmids. Amp, ampicillin. Name Strain PMM34 PMM66 PMM137 Plasmid pPET1 pRS04 pRS22 pRS23 pM26D pETD22 pRT11 Relevant characteristics Ref. ⁄ source Mycobacterium tuberculosis H37Rv fadD29::res–km–res, KmR M. tuberculosis H37Rv fadD29::res Mycobacterium bovis BCG Pasteur fadD26 This study This study This study pMIP12H containing pks15 ⁄ 1 from M. bovis BCG, HygR pMV361 containing fadD29 from M. tuberculosis, KmR pMV361 containing Rv2949c from M. tuberculosis, KmR pMV361 containing fadD29 and Rv2949c from M. tuberculosis, KmR pMV361e containing fadD26 from M. bovis BCG, KmR pET26b containing fadD22 from M. tuberculosis H37Rv, KmR pET21a containing fadD29 from M. tuberculosis H37Rv, AmpR [7] This study This study This study This study This study Gift from R. S. Gokhale FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS 2723 Phthiocerol dimycocerosates in M. tuberculosis R. Siméone et al. visualized by spraying the plates, respectively, with 10% phosphomolybdic acid in ethanol, and with a 0.2% (w ⁄ v) anthrone solution in concentrated H2SO4, followed by heating. For MALDI-TOF MS analyses, DIMs and PGLs were purified by preparative TLC as described above, and recovered by scraping silica gel from the plates. Production of DIMs in the various M. tuberculosis and M. bovis BCG strains was also visualized by radiolabeling newly synthesized lipids. Four microcuries (1.48 · 105 Bq) of sodium [1-14C]propionate (specific activity, 2.03 · 1012 BqÆmol)1; ICN Biomedicals, Irvine, CA, USA) was added to 8 mL of log-phase cultures and incubated for 24 h. Lipids were then extracted and separated by TLC as described above, and labeled compounds were quantified using a PhosphorImager (Typhoon Trio; GE Healthcare, Saclay, France). MALDI-TOF MS MALDI-TOF MS was performed using a voyager DE-STR MALDI-TOF instrument (PerSeptive Biosystems, Framingham, MA, USA) equipped with a pulse nitrogen laser emitting at 337 nm, as previously described [17,23]. Production of FadD22 and FadD29 and enzymatic assays for acyl-AMP ⁄ p-HB-AMP formation Cultures of E. coli BL21(DE3) transformed with plasmid pETD22 or plasmid pRT11 were grown for 72 h at 20 C in 250 mL of autoinducing medium [24]. Cells were collected by centrifugation (10 min, 5000 g), washed in buffer A (50 mm Tris, pH 8, 300 mm NaCl, 10% glycerol), resuspended in 25 mL of lysis buffer (0.75 mgÆmL)1 lysozyme, 50 mm Tris, pH 8, 300 mm NaCl, 40 mm imidazole, 1 mm phenylmethanesulfonyl fluoride), and stored at )80 C. The bacterial suspension was thawed at room temperature, and sonicated in ice with a Vibra cell apparatus (Fisher Bioblock Scientific, Illkirch, France). After centrifugation (30 min, 20 000 g), the cleared lysate was filtered with a 0.2 lm membrane (Millipore, Billerica, MA, USA) and applied to a HisTrap HP (1 mL) column equilibrated with 40 mm imidazole in buffer B (50 mm Tris, pH 8, 300 mm NaCl, 0.2 mm phenylmethanesulfonyl fluoride). After a step wash at 60 mm imidazole, FadD22 and FadD29 were eluted at 250 mm imidazole in buffer B. Fractions corresponding to the eluted protein were concentrated by ultrafiltration before storage at )20 C in the presence of glycerol (50%, v ⁄ v). The standard FAAL reaction mixture included 100 mm Tris ⁄ HCl (pH 7), 100 lm [14C]lauric acid (specific activity, 2.0 · 1012 BqÆmol)1; ARC, St Louis, MO, USA), 2 mm ATP, 8 mm MgCl2 and 2 lg of protein in a 15 lL reaction volume. The reaction, performed for 2 h at 30 C, was initiated by the addition of the enzyme, terminated by the addition of 5% acetic acid, and spotted onto silica gel TLC plates (G60). Radiolabeled products were redissolved in 2724 butan-1-ol ⁄ acetic acid ⁄ water (80 : 25 : 40) at room temperature. For p-HB-AMP ligase activity determination, the reaction mixture contained 75 mm Mes (pH 6.5), 0.5 mm MgCl2, 1 mm Tris(2-carboxyethyl)phosphine, 50 lm [14C] p-HBA (specific activity, 2.0 · 1012 BqÆmol)1; ARC), 2 mm ATP, and 2 lg of protein. Reactions were initiated by the addition of the enzyme, incubated for 2 h at 30 C, and stopped in ice. Product formation was analyzed by using silica gel TLC plates (G60) developed in ethyl acetate ⁄ isopropyl alcohol ⁄ acetic acid ⁄ water (70 : 20 : 25 : 40, v ⁄ v). TLC plates were exposed to phosphor imaging plates and quantificated by use of a PhosphorImager. Acknowledgements We are grateful to R. S. Gokhale (National Institute of Immunology, India) for providing vector pRT11, and F. Laval (IPBS, Toulouse) for her valuable assistance with MS. R. Siméone and M. Léger are recipients of fellowships from the European Commission and from the French Ministry of Teaching and Scientific Research. This work was supported by the Agence Nationale de la Recherche (No. ANR-06-MIME-032). References 1 Daffé M & Draper P (1998) The envelope layers of mycobacteria with reference to their pathogenicity. Adv Microb Physiol 39, 131–203. 2 Astarie-Dequeker C, Le Guyader L, Malaga W, Seaphanh FK, Chalut C, Lopez A & Guilhot C (2009) Phthiocerol dimycocerosates of M. tuberculosis participate in macrophage invasion by inducing changes in the organization of plasma membrane lipids. PLoS Pathog 5, e1000289. 3 Cox JS, Chen B, McNeil M & Jacobs WR Jr (1999) Complex lipid determines tissue-specific replication of Mycobacterium tuberculosis in mice. Nature 402, 79–83. 4 Rousseau C, Winter N, Pivert E, Bordat Y, Neyrolles O, Ave P, Huerre M, Gicquel B & Jackson M (2004) Production of phthiocerol dimycocerosates protects Mycobacterium tuberculosis from the cidal activity of reactive nitrogen intermediates produced by macrophages and modulates the early immune response to infection. Cell Microbiol 6, 277–287. 5 Camacho LR, Constant P, Raynaud C, Lanéelle MA, Triccas JA, Gicquel B, Daffé M & Guilhot C (2001) Analysis of the phthiocerol dimycocerosate locus of Mycobacterium tuberculosis. Evidence that this lipid is involved in the cell wall permeability barrier. J Biol Chem 276, 19845–19854. 6 Daffé M & Lanéelle MA (1988) Distribution of phthiocerol diester, phenolic mycosides and related FEBS Journal 277 (2010) 2715–2725 ª 2010 The Authors Journal compilation ª 2010 FEBS