Definition of the First Mannosylation Step in Phosphatidylinositol Mannoside Synthesis

We examined the function of the pimA(Rv2610c) gene, located in the vicinity of the phosphatidylinositol synthase gene in the genomes of Mycobacterium tuberculosisand Mycobacterium smegmatis, which encodes a putative mannosyltransferase involved in the early steps of phosphatidylinositol mannoside synthesis. A cell-free assay was developed in which membranes from M. smegmatis overexpressing the pimA gene incorporate mannose from GDP-[14C]Man into di- and tri-acylated phosphatidylinositol mono-mannosides. Moreover, crude extracts from Escherichia coli producing a recombinant PimA protein synthesized diacylated phosphatidylinositol mono-mannoside from GDP-[14C]Man and bovine phosphatidylinositol. To determine whether PimA is an essential enzyme of mycobacteria, we constructed a pimA conditional mutant of M. smegmatis. The ability of this mutant to synthesize the PimA mannosyltransferase was dependent on the presence of a functional copy of the pimA gene carried on a temperature-sensitive rescue plasmid. We demonstrate here that the pimA mutant is unable to grow at the higher temperature at which the rescue plasmid is lost. Thus, the synthesis of phosphatidylinositol mono-mannosides and derived higher phosphatidylinositol mannosides in M. smegmatisappears to be dependent on PimA and essential for growth. This work provides the first direct evidence of the essentiality of phosphatidylinositol mannosides for the growth of mycobacteria.

Phosphatidylinositol (PI) 1 and phosphatidylinositol mannosides (PIMs) 2 are the prominent and most distinguishable phospholipids of mycobacteria. They also provide the lipid anchor of two lipoglycans, lipomannan and lipoarabinomannan, the latter being an important modulator of the immune response in the course of tuberculosis and leprosy (1-3) as well as a key ligand in the interactions between Mycobacterium tuberculosis and phagocytic cells (4 -7). In Mycobacterium bovis BCG (Bacille de Calmette Guerin), PI and PIMs represent as much as 56% of all phospholipids in the cell wall and 37% of those in the cytoplasmic membrane and are, thus, regarded as important structural components acting as "cementing substances" for the cell wall skeleton (8). In support of this assumption, the synthesis of PI was recently shown to be essential for growth of Mycobacterium smegmatis (9).
Although little is known about the biosynthesis of PIMs, lipomannan, and lipoarabinomannan, structural similarities based on a conserved glycosylated phosphatidylinositol anchor point to a metabolic relationship (10 -14). Studies begun more than 30 years ago provide evidence that the early steps of PIM synthesis start with the transfer of a mannose residue from GDP-Man to the 2-position of the myo-inositol ring of PI to form phosphatidylinositol monomannosides (PIM 1 ). This step is followed by the transfer of another mannose residue (Man) to the 6-position of myo-inositol to form phosphatidylinositol dimannosides (PIM 2 ) (15)(16)(17). From PIM 2 , it is proposed that the Man residue in the 6-position of myo-inositol is further glycosylated with Man and then with Ara to form the higher forms of PIMs (PIM 3 -PIM 6 ) and the highly branched lipoglycans lipomannan and lipoarabinomannan through reactions probably involving many different mannosyl-and arabinosyltransferases as well as nucleoside diphosphate-and polyprenyl phosphate-based sugar donors (14). Recently, the pimB gene of M. tuberculosis was characterized as encoding a ␣-D-mannose-␣(136)-phosphatidyl-myo-inositol-monomannoside transferase that mediates the transfer of Man from GDP-Man to tri-acylated PIM 1 (Ac 3 PIM 1 ) to form tri-acylated PIM 2 (Ac 3 PIM 2 ) (18). The amino acid sequence of PimB contains the motif EXF(G/C)XXXXE found in bacterial retaining ␣-mannosyltransferases that catalyze the formation of glycosidic bonds using GDP-Man as the sugar donor (19). This motif, which is proposed to be involved in the binding of GDP-Man, is found in four other M. tuberculosispredicted proteins (20), among which is Rv2610c (amino acid residues 274 -282). Interestingly, Rv2610c is the fourth gene of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ ‡ To whom correspondence should be addressed. Tel.: 33-1-45-68-88-77; Fax: 33-1-45-68-88-43; E-mail: mjackson@pasteur.fr. 1 The abbreviations used are: PI, phosphatidyl-myo-inositol; ORF, open reading frame; LB, Luria Bertani culture medium, PIM, phosphatidyl-myo-inositol mannoside; kb, kilobase(s); Km, kanamycin; Str, streptomycin; Hyg, hygromycin; Suc, sucrose; Km R , kanamycin-resistant; Str R , streptomycin-resistant; Hyg R , hygromycin-resistant; Suc R , sucrose-resistant; Ara, arabinose; MALDI, matrix-assisted laser desorption/ionization; TOF, time of flight; C 16 , palmitate; C 19 , tuberculostearate (10-methyloctadecanoate). 2 PIM is used to describe the global family of PIMs that carries one to four fatty acids and one to six mannose residues. In Ac x PIM y , x refers to the total number of acyl groups, including those attached to the glycerol (the diacylglycerol substituent), and y refers to the number of mannose residues; e.g. Ac 2 PIM 1 corresponds to the diacylated form of the phosphatidylinositol mono-mannoside PIM 1 . This nomenclature requires careful enunciation, since inherent in the abbreviation for phosphatidyl-myo-inositol, PI, is the diacylglycerol unit. a cluster of five ORFs (20) potentially organized as a single transcriptional unit (9) and likely to be involved in the synthesis of PIMs. The first ORF of this cluster (Rv2613c) encodes a protein of unknown function. The second ORF encodes the phosphatidylinositol synthase PgsA characterized earlier (9). The third ORF (Rv2611c) encodes a protein with similarities to bacterial acyltransferases, and the fourth and fifth ORF encode, respectively, the putative ␣-mannosyltransferase (Rv2610c) and a putative GDP-mannose hydrolase (Rv2609c) carrying a MutT domain signature (PS00893) (21). This genetic organization suggested that Rv2610c might encode a ␣-mannosyltransferase involved in the very early steps of PIM synthesis (9). Rv2610c shares sequence similarity to many bacterial glycosyltransferases, is present in all the mycobacterial genomes sequenced so far, and has a homolog in Streptomyces coelicolor (51% identity on a 375 amino acid overlap), an actinomycete that shares with mycobacteria the ability to synthesize PIMs (22).
In this report, evidence is provided that Rv2610c is the ␣-D-mannose-␣(132)-phosphatidyl-myo-inositol transferase responsible for the formation of Ac 2 PIM 1 from GDP-Man and PI. Based on the name recently given to the mannosyltransferase responsible for the synthesis of Ac 3 PIM 2 from Ac 3 PIM 1 (PimB) (18), we name the Rv2610c enzyme PimA. Through the construction and analysis of a pimA conditional mutant of M. smegmatis, we demonstrate that PimA is essential for mycobacterial growth.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-Escherichia coli XL1blue, the strain used in this study for cloning experiments, and E. coli BL21(DE3)pLysS were routinely propagated in Luria Bertani (LB) broth (Difco) at 37°C. M. smegmatis strain mc 2 155 (23) was routinely grown at 30, 37, or 42°C in LB broth supplemented with 0.05% Tween 80. LB medium was used as the solid medium for all bacteria. Antibiotics were added at the following concentrations: ampicillin, 100 g/ml; chloramphenicol, 34 g/ml; kanamycin, 20 g/ml; hygromycin, 50 g/ ml; streptomycin, 20 g/ml. When required, 10% sucrose was added to the solid medium.
Cloning Procedures, Mycobacterial Genomic DNA Extraction, and Southern Analysis-For preparation of electrocompetent cells, E. coli XL1-blue cells grown in LB broth were washed 2 times with distilled water, 1 time in 10% glycerol, and finally, resuspended in 10% glycerol. The same procedure was used for the preparation of M. smegmatis electrocompetent cells except that 0.05% Tween 80 was added to all washing solutions. Aliquots of electrocompetent cells were transformed using a Gene Pulser unit (Bio-Rad) with a single pulse (2.5 kV; 25 microfarads; 200 ohms). Purification of DNA restriction fragments and PCR fragments were performed using the QIAquick gel extraction kit and QIAquick PCR purification kit (Qiagen, Chatsworth, CA). Plasmids were isolated from E. coli XL1-blue using the QIAprep miniprep kit (Qiagen). Mycobacterial genomic DNA was isolated as follows. Cells from a 5-ml overnight culture were pelleted by centrifugation at 3500 rpm for 15 min. The pellet was resuspended in 250 l of solution I (25% sucrose, 50 mM Tris-HCl, pH 8.0, 50 mM EDTA, 500 g/ml lysozyme) and incubated overnight at 37°C under agitation. The next day, 250 l of solution II (100 mM Tris-HCl, pH 8.0, 1% SDS, 400 g/ml proteinase K) were added, and the reaction mixture was incubated at 55°C for 4 h. DNA was extracted twice with phenol and chloroform and concentrated by ethanol precipitation. Molecular cloning and restriction endonuclease digestions were performed by standard techniques according to the manufacturer's recommendations. Labeling of DNA probes with [␣-32 P]dCTP and Southern blot analyses were performed as described (24). The M. smegmatis genomic DNA sequences used in this study were obtained from the TIGR Center (www.tigr.org). Sequences were processed using the DNA Strider program (Commissariat à l'Energie Atomique, Gif-sur-Yvette, France).
Overexpression of the M. smegmatis pimA Gene in M. smegmatis-Standard PCR strategies with Taq DNA polymerase (Applied Biosystems, Roche Molecular Biochemicals) were used to amplify the M. smegmatis pimA gene. PCR amplification consisted of 1 denaturation cycle (95°C, 6 min) followed by 40 cycles of denaturation (95°C, 1 min), annealing (60°C, 1 min) and primer extension (72°C, 1.5 min), and a final extension at 72°C for 10 min. The plasmid pCGpisB (see below) was used as the DNA template, and the primers were ManT1 (5Јccaccaacatatgcgtatcgggatggtctgccc-3Ј) and ManT2 (5Ј-cccaagcttgaccgattctccggccgtctcg-3Ј). The primers were designed to generate a PCR product corresponding to the entire pimA gene devoid of its stop codon and harboring NdeI and HindIII restriction sites (underlined in the primer sequences), enabling direct cloning into the pVV16 expression vector (9). pVV16 harbors a kanamycin and a hygromycin resistance marker and allows genes to be constitutively expressed under the control of the hsp60 transcription and translation signals. Recombinant proteins produced with this system carry a six-histidine tag at their C terminus. M. smegmatis mc 2 155 was transformed with the resulting expression vector, pVVpimA, and transformants were selected on LB-Km-Hyg plates. The production of recombinant PimA protein in M. smegmatis was analyzed by SDS-PAGE on a 12% gel followed by immunoblotting using a nitrocellulose transfer membrane Hybond C (Amersham Biosciences) with a mouse monoclonal anti-His antibody (Penta-His antibody, Qiagen) diluted 1:1000. The secondary antibody was a horseradish peroxidase-conjugated sheep anti-mouse IgG antibody (Amersham Biosciences) was used at a 1:10,000 dilution. Bound antibodies were detected using the ECL system (Amersham Biosciences). M. smegmatis crude extracts were prepared by harvesting cultures of the recombinant strains (A 600 nm ϭ 1.5), suspending them in phosphate-buffered saline, subjecting them to probe sonication at 4°C for 5 min in the form of 5 ϫ 60-s pulses with 60 s cooling intervals between pulses, and removing the unbroken cells and bacterial debris by centrifugation of the sonicate at 10,000 ϫ g for 15 min.
Drug Sensitivity Assays-LB-agar medium containing isoniazid, chloramphenicol, or ampicillin in dilution series was added to the wells of six-well plates. These plates were inoculated with appropriate dilutions of mc 2 155 or mc 2 155/pVVpimA cultures and incubated for 3-4 days at 37°C. 99% inhibition of the bacterial growth was determined as the minimal inhibitory concentration of the drug.
Expression of the M. smegmatis pimA Gene in E. coli-Recombinant PimA protein was produced in E. coli BL21(DE3)pLysS using the pET14b expression system (Novagen, Madison, WI). The M. smegmatis pimA gene was amplified using Taq DNA polymerase (Applied Biosystems, Roche Molecular Biochemicals), pCGpisB (see below) as the DNA template, and the primers Man1 (5Ј-cggcgggcatatgcgtatcgggatggtctgc-3Ј) and Man2 (5Ј-cccggatcctcagaccgattctccggccgt-3Ј). PCR amplification consisted of one denaturation cycle (95°C, 6 min) followed by 40 cycles of denaturation (95°C, 1 min), annealing (64°C, 1 min), and primer extension (72°C, 1.5 min) and a final extension at 72°C for 10 min. The primers were designed to generate a PCR product corresponding to the entire pimA gene and harboring NdeI and BamHI restriction sites (underlined in the primer sequences), enabling direct cloning into the pET14b expression vector (Novagen). The resulting expression vector was named pETpimA. It allows for the expression of the pimA gene under control of the strong bacteriophage T7 transcription and translation signals and for the production of an N-terminal six-histidine tagged PimA recombinant protein. The production of recombinant PimA protein in E. coli BL21/pETpimA transformants was induced by the addition of 0.4 mM isopropyl-␤-D-thiogalactopyranoside to the culture medium with incubation at 37°C for 3 h. The production of PimA protein was then analyzed by immunoblotting with a mouse monoclonal anti-His antibody (Penta-His antibody, Qiagen) as described for the recombinant PimA protein produced in M. smegmatis.
Cell-free Assay for Mannophospholipid Synthesis in M. smegmatis Using GDP-[ 14  H]inositol-labeled lipids from mc 2 155/pVV16 and mc 2 155/pVVpimA. Lipid extracts (200,000 dpm) were loaded to TLC plates and developed in two dimensions as described under "Experimental Procedures." b, MALDI-MS analysis of the total lipids from mc 2 155/pVV16 and mc 2 155/pVVpimA. Cold lipids from mc 2 155/pVV16 and mc 2 155/pVVpimA were extracted and subjected to MALDI-MS analysis in the negative ion mode as described under "Experimental Procedures." Peaks observed are m/z 851.6, Ac 2 PI with C 16  mixtures contained 4 mg of membrane proteins from mc 2 155/pVV16 or mc 2 155/pVVpimA, 20 M GDP-Man (Sigma), 10 mM MgCl 2 , 62.5 M ATP, and buffer A, pH 7.45, in the final volume of 1 ml. The total lipids from six reactions were extracted as described above, combined and analyzed by MALDI-MS. For the individual characterization of mannolipid products, the non-radioactive mannolipids were isolated by preparative TLC using the radiolabeled mannolipids as markers. After autoradiography, the relevant regions of the TLC were scraped off, extracted with CHCl 3 /CH 3 OH (2:1), and subjected to MALDI-MS analysis. The same amounts of mannolipids were produced by membrane fractions of mc 2 155/pVV16 and mc 2 155/pVVpimA in the absence or presence of ATP in the reaction mixture, suggesting that ATP is not required in the assay.
Cell-free Assay for Mannophospholipid Synthesis in E. coli-E. coli BL21/pET14b and E. coli BL21/pETpimA (4 g wet weight) were suspended in 4 ml of buffer A, pH 7.45, and subjected to probe sonication for 3 min 20 s in the form of 20 ϫ 10-s pulses with 60-s cooling intervals between pulses. The sonicate was centrifuged for 20 min at 10,000 ϫ g to remove the unbroken cells and bacterial debris, and the resulting supernatant (E. coli crude extract) was kept frozen in small aliquots at Ϫ20°C and then used directly in cell-free assays as the sole source of enzymes. Protein concentrations in the E. coli BL21/pET14b and E. coli BL21/pETpimA extracts were 20 and 55 mg/ml, respectively.
Reaction Analytical Procedures-Lipids from labeled and non-labeled cells were extracted by two consecutive overnight extractions in CHCl 3 / CH 3 OH (2:1) followed by one overnight extraction in CHCl 3 /CH 3 OH (1:2). The CHCl 3 /CH 3 OH extracts were combined, Folch-washed (CHCl 3 /CH 3 OH/H 2 O (4:2:1)), and dried before reconstituting in CHCl 3 / CH 3 OH (2:1) for analysis by TLC. Characterization of the various PIMs followed earlier work (9,15,18) and was based on one-and twodimensional thin-layer chromatographic patterns, co-migration of labeled PIM products with authentic PIM standards, and mass spectrometry analysis of total or purified lipids. The fatty acids of the PIMs are primarily C 16 (palmitate) and C 19 (tuberculostearate) (15). TLC was conducted on aluminum-backed plates of silica gel 60 F 254 (Merck). The solvent system used was CHCl 3 /CH 3  The lipids from six reactions were extracted and combined, and mannolipids 1 and 2 were isolated by preparative TLC using the radiolabeled mannolipids as markers. MALDI-MS was operated in the negative ion mode. Peaks observed are m/z 1013.6, Ac 2 PIM 1 with C 16  , v/v) in the second dimension. An ␣-naphthol spray (1% ␣-naphthol in ethanol) and a cupric sulfate spray (10% CuSO 4 in a 8% phosphoric acid solution) were used to detect carbohydratecontaining lipids and all organic compounds, respectively. Autoradiograms were obtained by exposing chromatograms to Kodak BIOMAX MR films at Ϫ70°C for 1-7 days. Relevant spots were scraped off for scintillation counting. For MALDI-MS analysis of the phospholipids containing fraction, total lipid extracts were suspended in CHCl 3 and washed twice with an equal volume of water. The organic phase was then brought to dryness, solubilized in acetone, and allowed to precipitate overnight at 4°C. The suspension was then centrifuged at 4°C for 15 min (3,000 ϫ g). The supernatant was removed, and the precipitate was resuspended in CHCl 3 and analyzed in MALDI-TOF-MS.
Sample Preparation and MALDI-TOF Mass Spectrometry-Analysis by MALDI-TOF-MS was carried out on a Voyager DE-STR (PerSeptive Biosystems, Framingham, MA) using the reflectron mode. Ionization was effected by irradiation with pulsed UV light (337 nm) from an N 2 laser. PIMs were analyzed by the instrument operating at 20 kV in the negative ion mode using an extraction delay time set at 200 ns. Typically, spectra from 100 to 250 laser shots were summed to obtain the final spectrum. All of the samples were prepared for MALDI analysis using the on-probe sample cleanup procedure with cation-exchange resin. The 2-(4-hydroxyphenylazo)-benzoic acid (HABA) matrix was used at a concentration of ϳ10 mg/ml in ethanol/water (1:1 v/v). Typically, 0.5 l of PIM sample (10 g) in a CHCl 3 /CH 3 OH/H 2 O solution and 0.5 l of the matrix solution, containing ϳ5-10 cation exchange beads, were deposited on the target, mixed with a micropipet, and dried under a gentle stream of warm air. The measurements were externally calibrated at two points with PIMs.
Construction of the M. smegmatis pimA Conditional Mutant-The essentiality of the pimA gene in M. smegmatis was investigated using a two-step homologous recombination procedure. The method relies upon the use of a suicide vector harboring the counterselectable marker sacB and a kanamycin cassette-disrupted copy of the gene of interest. In the first step of the experiment, a single crossover strain is isolated. This strain contains the sacB gene and should be sensitive to sucrose. In the second step of the experiment, a culture of the single crossover strain is plated onto sucrose-Km plates to select for clones that undergo a second intrachromosomal crossover, leading to the excision of the body of the vector and to allelic replacement. In the case in which the gene of interest is essential, allelic replacement at the second step of the experiment should be achievable only in the presence of a rescue copy of this gene provided on a replicative or integrative vector.
The M. smegmatis pimA gene and flanking regions were excised from the plasmid pUCpgsA.Sm (9) on a 3.7-kb BamHI restriction fragment and inserted at the BamHI site of pACYC184 (New England Biolabs, Inc.), yielding plasmid pACYCpimA. A disrupted allele of the pimA gene, pimA::Km, was then constructed by cloning the kanamycin resistance cassette from pUC4K (Amersham Biosciences) carried on a 1.2-kb HincII restriction fragment into the KpnI-cut and blunt-ended pACY-CpimA. pimA::Km was then excised from the resulting plasmid on a 4.9-kb BamHI restriction fragment, blunt-ended, and inserted at the SmaI site of pXYL4 (a pBlueScript derivative carrying the xylE reporter gene) (25), yielding plasmid pX4pimAK. Finally, pJQpimA, the construct used for allelic replacement, was obtained by transferring a 5.9-kb BamHI fragment from pX4pimAK, containing pimA::Km and xylE, into the BamHI-cut pJQ200, an E. coli cloning vector carrying the counterselectable marker sacB (26).
pCG76, a Mycobacterium/E. coli shuttle plasmid harboring a mycobacterial temperature-sensitive origin of replication and a streptomycin resistance cassette (27), was used as the rescue plasmid to carry a functional copy of the pimA gene in the M. smegmatis pimA conditional Cold cell-free assays using crude extracts from E. coli BL21/pETpimA were performed as described under "Experimental Procedures." The lipids from 20 reactions were extracted, combined, and subjected to MALDI-MS analysis in the negative ion mode. Only the major species of each family are marked. The peak at m/z 885.6 is attributed to Ac 2 PI with C 18 /C 20:4 . Peaks at m/z 887.6 could then be tentatively attributed to Ac 2 PI with C 18 /C 20:3 and m/z 861.6 and 863.6 to Ac 2 PI with 2C 18:1 and C 18 /C 18:1 respectively. mutant. pCGpisB, one of the rescue plasmids used in this study, was constructed by inserting the 3.7-kb BamHI restriction fragment from pUCpgsA.Sm (9) at the BamHI site of pCG76. The 3.7-kb BamHI insert carries full-length wild type copies of the M. smegmatis pgsA, Rv2611c, and pimA (Rv2610c) genes.

Overexpression of the pimA Gene in M. smegmatis
Sub-cellular Localization of the PimA Protein-The Rv2610c gene from M. tuberculosis, which encodes a putative mannosyltransferase involved in the early steps of PIM synthesis, was renamed pimA. The pimA gene of M. smegmatis was PCRamplified and placed under control of the phsp60 promoter in the mycobacterial expression vector pVV16, yielding plasmid pVVpimA. Upon transformation of M. smegmatis mc 2 155 with this construct, colonies of mc 2 155/pVVpimA were obtained that exhibited an unusual glossy morphology instead of the dry morphology of wild type mc 2 155 or mc 2 155/pVV16 colonies. The growth rates of the mc 2 155/pVVpimA and mc 2 155/pVV16 recombinant strains were identical in LB-Km-Hyg-Tween 80 broth at 37°C, although the strain overexpressing pimA showed an increased tendency to clump (data not shown). Production of recombinant PimA protein in mc 2 155/pVVpimA was checked by Western blot using a mouse monoclonal anti-His antibody. Large quantities of PimA recombinant protein of the expected size (ϳ40 kDa) were detected in crude extracts of mc 2 155/pVVpimA and found to be associated to the membrane and cytosol fractions (data not shown). The association of PimA with the membrane fraction is consistent with the prediction of at least one putative transmembrane segment from the amino acids 275-298 by the TMpred (www.ch.embnet.org) and DAS (www.sbc.su.se/ϳmiklos/DAS) transmembrane prediction programs. The uneven distribution of basic amino acid residues responsible for a high predicted pI of the N-terminal half of the protein (theoretical pI ϭ 9.55 from residues 1 to 202) as compared with a theoretical pI of 5.10 for the C-terminal half of the protein (residues 203-378), may also reflect the ability of some N-terminal domains of PimA to interact with anionic phospholipids of the membrane. Finally, the association of PimA with the membrane fraction is consistent with the observed co-localization of the mannosyltransferase activity with the membrane fraction, as shown below. The important amounts of non-active recombinant PimA protein in the cytosol fraction of mc 2 155/pVVpimA are probably the results of overexpression. Overexpression of pimA Alters the PIM Composition of M. smegmatis-The PIM composition of the mc 2 155/pVV16 and mc 2 155/pVVpimA strains was analyzed by metabolic labeling with myo- [2-3 H]inositol. Overexpression of pimA in M. smegmatis resulted in an increased production of PIM 2 , particularly Ac 3 PIM 2 , relative to PI (Fig. 1a). The spots corresponding to PI, Ac 4 PIM 2 , and Ac 3 PIM 2 (Fig. 1a) were scraped off and counted for radioactivity. The ratio of PI to the two forms of PIM 2 combined was 1:2.2 in the control strain mc 2 155/pVV16 and 1:13.6 in the overproducing strain mc 2 155/pVVpimA. Changes in the PI to PIM 2 ratio were further confirmed by MALDI-MS analysis in the negative ion mode of the phospholipidcontaining fraction obtained after chloroform/water partition and acetone precipitation (Fig. 1b) Overexpression of pimA Alters M. smegmatis Sensitivity to Ampicillin-The low permeability of the mycobacterial cell envelope and, subsequently, the high intrinsic resistance of mycobacteria to chemotherapeutic agents are believed to be a result of the unusual structure and composition of the cell envelope (28). Because PIMs represent major components of the cell envelope, we investigated whether alterations in the PIM composition of mc 2 155/pVVpimA affected its sensitivity to drugs. The minimal inhibitory concentrations of three antibiotics against mc 2 155/pVVpimA and mc 2 155/pVV16 were measured. Both strains exhibited identical resistance to isoniazid (3 g/ml) and chloramphenicol (30 g/ml) (hydrophilic and hydrophobic drugs, respectively), but mc 2 155/pVVpimA showed higher resistance to ampicillin, a hydrophilic ␤-lactam (minimal inhibitory concentration ϭ 300 -400 g/ml as compared with 100 g/ml for the control strain). This result is in good agreement with the observation by Parish and co-workers (29) that a M. smegmatis mutant with decreased amounts of PIM 2 in its envelope was more sensitive to ampicillin.

Overproduction of PimA in M. smegmatis Stimulates the Formation of Ac 2 PIM 1 and Ac 3 PIM 1 in Cell-free Assays
An enzymatic system associated to the membrane fraction of Mycobacterium phlei and capable of forming PIM 1 from GDP-Man was initially described by Hill and Ballou (16). GDP-Man was shown to be the only effective mannose donor in this system. In the present study, the mannosyltransferase activities of mc 2 155/pVV16 and mc 2 155/pVVpimA were compared in a cell-free assay based on that described for M. phlei. In this assay, a whole array of mannose-containing lipids, including polyprenol-based mannolipids and PIMs, are synthesized. Changes in the lipid profiles of the PimA-overproducing strain as compared with the control strain would suggest an involvement of PimA in a metabolic pathway leading to mannosecontaining lipids. Because the recombinant PimA protein was found in the cytosol and membranes of the overproducing strain mc 2 155/pVVpimA, we first tested these two fractions independently or in combination for mannosyltransferase activity. The radioactive profiles of the lipids obtained after incubation of membranes with GDP-[ 14 C]Man showed a clear accumulation of two mannolipids (mannolipids 1 and 2) migrating in the PIM region of the TLC in the case of mc 2 155/pV-VpimA (Fig. 2, lane 2). The reactions performed with the cytosol of the control and overproducing strains yielded identical radioactive lipid profiles (Fig. 2, lanes 3 and 4). Almost no mannolipid 1 and no mannolipid 2 could be detected when using the cytosol fraction, even when mycobacterial PI (ϳ15 M) was added to the reaction mixture (data not shown). The formation of small amounts of polyprenyl phosphomannose and Ac 3 PIM 2 in the cytosol fraction is likely the result of a contamination of this fraction with traces of membrane. Combining the membrane and cytosol fractions did not increase significantly the synthesis of mannolipid 1, although slightly higher amounts of mannolipid 2 were detected in both the control and overexpressing strains (Fig. 2, lanes 5 and 6). Thus, despite of the fact that PimA is present in the cytosol of M. smegmatis mc 2 155/pVVpimA, the mannosyltransferase activity of this protein is associated only with the membrane fraction. Initial characterization of the two accumulated mannolipids confirmed that they are mild-acid stable and mild-alkali labile, indicating that they are members of the PIM family (data not shown). Further analysis by MALDI-MS of the total lipids from reaction mixtures in which radioactive GDP-[ 14 C]Man was replaced by cold GDP-Man revealed the presence of two compounds corresponding to Ac 2 PIM 1 ([M-H] Ϫ , m/z ϭ 1013.6) and Ac 3 PIM 1 ([M-H] Ϫ , m/z ϭ 1251.9) in mc 2 155/pVVpimA that were not detected in mc 2 155/pVV16 (data not shown). To characterize the mannolipids 1 and 2, non-radioactive products from cold reactions were partially purified by preparative TLC and subjected to MALDI-MS analysis. The mass spectrum for the mannolipid 1-containing sample shows the presence of the major product at m/z ϭ 1013.6, corresponding to Ac 2 PIM 1 with C 16 /C 19 (Fig. 3a), contaminated by traces of co-purified PIMs (Ac 2 PIM 2 , Ac 3 PIM 2 , and Ac 3 PIM 3 ). Mannolipid 2-containing sample afforded predominant deprotonated molecular ions at m/z ϭ 1251.9, corresponding to Ac 3 PIM 1 with 2C 16 /C 19 (Fig.  3b). Therefore, the major radiolabeled product of the reaction catalyzed by PimA in M. smegmatis is Ac 2 PIM 1 . Part of this product is presumably acylated into Ac 3 PIM 1 .

PimA Catalyzes the Formation of Ac 2 PIM 1 in E. coli Cell-free Assays
The function of the PimA protein was further studied in crude extracts of E. coli BL21/pETpimA overexpressing pimA. Because E. coli does not produce any PI, no background mannosyltransferase and acyltransferase activities on PI or derived products were expected in these extracts. In the reaction mixture, commercial bovine PI (carrying primarily stearate (C 18:0 ) and arachidonate (C 20:4 ) fatty acyl chains) and GDP-[ 14 C]Man served as the substrates for the predicted reaction, PI ϩ GDP-[ 14 C]Man 3 Ac 2 PI[ 14 C]M 1 ϩ GDP, catalyzed by the recombinant PimA protein. In the presence of crude extract from E. coli BL21/pETpimA but not in that of crude extract from BL21/ pET14b, a single product was formed (Fig. 4, lanes 3-4). Its synthesis was strictly dependent upon the addition of PI to the reaction mixture (Fig. 4, lanes 1-4). This product migrated with a similar Rf to Ac 2 PIM 1 , produced in the M. smegmatis cell-free assays (Fig 4, lanes 5 and 6). A slight shift in TLC mobility of this product can be attributed to the difference in acyl groups of mycobacterial PI (containing C 16 /C 19 ) compared with commercial bovine PI. MALDI-MS analysis of the total lipids from reaction mixtures in which radioactive GDP-[ 14 C]Man was replaced by cold GDP-Man showed that the only detected PIM corresponded to Ac 2 PIM 1 (with peaks at m/z ϭ 1025.7 and m/z ϭ 1049.7) in BL21/pETpimA (Fig. 5), which were not detected in BL21/pET14b. It is obvious that these compounds arose from PI (m/z ϭ 863.6; m/z ϭ 887.6) by the addition of a single hexose (i.e. mannose). Duplicity of the peaks on the spectrum is due to heterogeneity of commercial PI, which is claimed to contain primarily, but not exclusively, stearate and arachidonate fatty acyl chains.
Altogether, the results obtained from the M. smegmatis and E. coli cell-free assays provide evidence that the PimA enzyme of M. smegmatis catalyzes the transfer of the first Man residue from GDP-Man to the myo-inositol residue of PI, yielding Ac 2 PIM 1 . Because earlier work (16,17) clearly showed that the very first transfer of a Man residue onto PI in PIM synthesis occurred at the 2-position of myo-inositol, we therefore conclude that PimA is a ␣-D-mannose-␣(132)-phosphatidyl-myoinositol transferase.
pimA Is an Essential Gene of M. smegmatis Construction of a pimA Conditional Mutant of M. smegmatis-To address the question of the essentiality of the mannosyltransferase PimA in mycobacteria, we constructed a pimA conditional mutant of M. smegmatis. Essentially the same strategy was used to construct this mutant as was used to construct a pgsA conditional mutant (9). It uses a two-step homologous recombination procedure to achieve allelic exchange at the pimA locus (30) and a mycobacterial temperature-sensitive rescue plasmid to perform complementation experiments (27). A kanamycin-disrupted copy of the pimA gene, pimA::Km, and the xylE reporter gene were inserted into the sacB suicide vector pJQ200, yielding pJQpimA. pJQpimA was introduced into the wild type M. smegmatis mc 2 155 by electroporation, and kanamycin-resistant transformants were selected on LB-Km plates at 37°C. A Southern blot analysis performed on the DNA of 18 of these transformants indicated that 14 resulted from a single homologous recombination event at the pimA locus (data not shown). Two, mc 2 pJQpimA.1 and mc 2 pJQpimA.2, were selected for the subsequent steps of the experiment. The other four transformants, analyzed by Southern blotting, arose from illegitimate recombination. mc 2 pJQpimA.1 and mc 2 pJQpimA.2 were grown in LB-Km broth and then plated onto LB-Km-Suc to select for clones that had undergone a second intra-chromosomal crossover leading to the excision of the body of vector and to allelic replacement. Allelic exchange mutants are expected to carry the disrupted allele pimA::Km and to have lost the sacB and xylE genes carried by pJQpimA. Therefore, allelic exchange mutants should be resistant to kanamycin and to sucrose and remain white upon spraying with catechol (i.e. XylE negative). Spraying of thousands of kanamycin-sucrose resistant colonies with catechol revealed that none of them exhibited the expected phenotype for allelic exchange mutants. Instead, these clones had probably undergone some mutations in the sacB gene that conferred upon them resistance to sucrose. To investigate whether the failure to disrupt the pimA gene in this first experiment could be due to the essentiality of that gene, we next proceeded to the construction of a pimA conditional mutant.
For this purpose, the single crossover strain mc 2 pJQpimA.1 was transformed with the temperature-sensitive rescue plasmid pCGpisB or with the empty pCG76 vector. pCGpisB carries functional copies of the M. smegmatis pgsA, Rv2611c, and pimA genes. Transformants were selected on LB-Km-Str plates, grown in LB-Km-Str broth at permissive temperature (30°C) and finally plated onto LB-Km-Str-Suc plates at 30°C to select for allelic exchange mutants as described previously. The XylE phenotype of approximately 1000 Km R -Suc R -Str R colonies was tested for both types of transformants plated. No allelic exchange mutant (Km R -Suc R -Str R -XylE Ϫ colony) was found when mc 2 pJQpimA.1/pCG76 was plated, confirming our previous result obtained at 37°C with the non-transformed mc 2 pJQpimA.1 strain. In contrast, plating of mc 2 pJQpimA.1/ pCGpisB yielded a majority (69%) of Km R -Suc R -Str R colonies with a XylE-negative phenotype. Further analysis by Southern blot of a Km R -Suc R -Str R -XylE Ϫ colony revealed that it had undergone gene replacement at the pimA locus (Fig. 6a). pimA conditional mutants, thus, carry a non-functional pimA::Km gene on their chromosome and a functional pimA gene on a conditionally replicative vector.
In conclusion, allelic replacement at the chromosomal pimA locus of M. smegmatis was achievable only in the presence of a rescue plasmid carrying a functional copy of the pimA gene. To conclusively provide evidence that the inability to achieve gene inactivation at the pimA locus is attributable to the pimA gene alone and not to polar effects of the mutation affecting the expression of adjacent genes (and which could be complemented by the pCGpisB plasmid), another allelic replacement experiment was designed in which the single crossover strain mc 2 pJQpimA.1 transformed with either pVV16 or pVVpimA was plated onto LB-Km-Hyg-Suc plates at 37°C. Because of the presence of a kanamycin resistance marker on the pVV16 and pVVpimA vectors, both allelic exchange mutants and revertants were selected at this step. To distinguish between allelic replacement and reversion, Km R -Hyg R -Suc R -XylE Ϫ colonies were analyzed by PCR or Southern blot. In the case of mc 2 pJQpimA.1 transformed with the pVVpimA, 5 of the 14 clones (35.7%) tested were allelic exchange mutants (Fig. 6b). The remaining clones were revertants carrying a wild type allele of pimA on their chromosome. In the case of mc 2 pJQpimA.1 transformed with the empty pVV16 vector, the 10 clones tested were revertants (data not shown). Therefore, the expression of the only pimA gene from pVVpimA is sufficient to rescue a M. smegmatis pimA knock-out mutant. These results strongly suggest that the pimA gene is essential to mycobacteria.
Growth Characteristics of the pimA Conditional Mutant under Permissive and Non-permissive Conditions-To conclusively provide evidence that the pimA gene is essential to M. smegmatis, we investigated the ability of a pCGpisB-complemented pimA mutant of M. smegmatis (named strain MYC1572) to survive at 42°C, a temperature at which the pCGpisB vector is unable to replicate. The growth characteristics of MYC1572 and wild type mc 2 155 strain at 30 and 42°C are presented on Fig. 7, a and b, respectively. As expected, at 30°C, the temperature-sensitive pCGpisB vector replicates, and MYC1572 exhibited the same growth characteristics as the control strain mc 2 155. After a shift of temperature from 30 to 42°C, although the control strain continued to grow exponentially, the A 600 of the MYC1572 culture started to decline after 10 h, paralleling the loss of the temperature-sensitive rescue plasmid (Fig. 7b). Therefore, the pimA gene appears to be essential for mycobacterial growth. These data suggest that PimA has a unique function that cannot be compensated by any other mannosyltransferase in the bacteria under the different conditions tested.

Conclusions
This study provides evidence that the pimA gene encodes the ␣-mannosyltransferase involved in the transfer of the very first mannose residue from GDP-Man to the 2-position of the inositol moiety of phosphatidylinositol, leading to the synthesis of Ac 2 PIM 1 , the biosynthetic precursor of higher PIM, lipomannan, and lipoarabinomannan. The characterization of the pimA gene, located in the vicinity of the phosphatidylinositol synthase gene (pgsA), confirms the existence of a cluster of genes dedicated to the early steps of the synthesis of PIMs in all mycobacterial genomes sequenced so far (9). The remarkable conservation of this cluster of genes among Mycobacterium spp. is to be related to its essentiality for mycobacterial growth. Indeed, we previously demonstrated that phosphatidylinositol is an essential component of the mycobacterial cell envelope (9), and we now provide the first demonstration that PIMs are also essential for the growth of mycobacteria. The role of PIMs and derived lipoglycans (lipomannan and lipoarabinomannan) in the envelope may be structural, as suggested earlier (8). The PIM composition of the envelope may also have a profound impact on its permeability, as suggested by the increased resistance of a M. smegmatis strain overproducing phosphatidylinositol dimannosides (mc 2 155/pVVpimA) to the hydrophilic drug ampicillin. Finally, the essentiality of the PimA mannosyltransferase and its involvement in a biosynthetic pathway that is confined to Mycobacterium spp. and to a few other actinomycetes (22) makes it an attractive drug target for antituberculosis therapy.