The nonredundant roles of two4�-phosphopantetheinyl transferasesin vital processes of MycobacteriaChristian Chalut†‡, Laure Botella†, Celia de Sousa-D’Auria§, Christine Houssin§, and Christophe Guilhot†

†Institut de Pharmacologie et de Biologie Structurale, Centre National de la Recherche Scientifique and Universite P. Sabatier (Unite Mixte de Recherche5089), 205 Route de Narbonne, 31077 Toulouse Cedex, France; and §Laboratoire de Biotechnologie des Microorganismes d’Interet Industriel, Institut deGenetique et Microbiologie, Centre National de la Recherche Scientifique and Universite Paris Sud (Unite Mixte de Recherche 8621), Bat 409,91405 Orsay Cedex, France

Edited by E. Peter Greenberg, University of Washington School of Medicine, Seattle, WA, and approved April 14, 2006 (received for reviewDecember 23, 2005)

Mycobacterium tuberculosis contains >20 enzymes that requireactivation by transfer of the 4�-phosphopantetheine moiety of CoAonto a conserved serine residue, a posttranslational modificationcatalyzed by 4�-phosphopantetheinyl transferases (PPTases). Themodified proteins are involved in key metabolic processes such ascell envelope biogenesis and the production of virulence factors.We show that two PPTases conserved in all Mycobacterium spp.and in related genera activate two different subsets of proteinsand are not functionally redundant. One enzyme, AcpS, activatesthe two fatty acid synthase systems of mycobacteria, whereas theother PPTase, PptT, acts on type-I polyketide synthases and non-ribosomal peptide synthases, both of which are involved in thebiosynthesis of virulence factors. We demonstrate that both PPTa-ses are essential for Mycobacterium smegmatis viability and thatPptT is required for the survival of Mycobacterium bovis bacillusCalmette–Guerin. These enzymes are thus central to the biology ofmycobacteria and for mycobacterial pathogenesis and representpromising targets for new antituberculosis drugs.

fatty acid synthase � lipid metabolism � mycolic acids � polyketide

The cell envelope plays a major role in the physiology ofMycobacterium tuberculosis, the causative agent of tubercu-

losis in humans. This complex structure protects the bacteriumagainst degradation by host enzymes and acts as an impermeablebarrier against antibiotics and toxic molecules produced by thehost. It also contains components that actively facilitate uptakeof the bacterium and modulate host immune response (1). Themycobacterial envelope has a very high lipid content and con-tains lipids with unusual structures, such as the mycolic acids (2).These very long chain fatty acids are specific to suborderCorynebacterineae and are the major lipid constituents of the cellwall (1, 3). They are key structural components of the cellenvelope, and their biosynthesis pathway is targeted by isoniazid,one of the front-line antituberculosis drugs (4). In M. tuberculosisand defined slow-growing mycobacteria, mycolic acid-containing substances and a number of extractable lipids con-taining methyl-branched fatty acids such as phthiocerol dimy-cocerosates (DIM) and phenolic glycolipid (PGL-tb) are knownto contribute to their pathogenicity (5, 6, 7).

The genome of M. tuberculosis encodes �18 type-I polyketidesynthases (Pks) and two fatty acid synthase (Fas) systems (8).These enzymes endow M. tuberculosis with its unique ability toproduce an impressive variety of lipids of unusual structure. Forinstance, mycolate production requires two Fas systems and onePks, and the biosynthesis of DIM and PGL-tb requires Fas-I andseven Pks (9, 10). The acyl carrier protein (ACP) domains of Fasand Pks are only functional if converted from their inactiveapo-forms to their functional holo-forms by the covalent attach-ment of a 4�-phosphopantetheine (P-pant) group to the hydroxylgroup of an invariant serine residue (11, 12). This feature is

shared by another class of enzymes, the nonribosomal peptidesynthases (NRPS), which are involved in the production ofsiderophores in M. tuberculosis (13). This posttranslational mod-ification is catalyzed by 4�-phosphopantetheinyl transferases(PPTases), which transfer the P-pant group from CoA to theACP (14). Two genes encoding PPTases, acpS and pptT, wereidentified in the genome of M. tuberculosis H37Rv (8, 15).However, the role of each PPTase and the putative redundancyof these proteins had not been investigated in mycobacteria. Inparticular, despite the importance of the Fas-I system and type-IPks for the biology of M. tuberculosis, no data concerning theposttranslational modification of these enzymes have been re-ported in mycobacteria. These issues are crucial to understand-ing lipid biosynthesis in M. tuberculosis.

In this article, we report that orthologs of the two M. tuber-culosis PPTases are found in other Corynebacterineae species.We demonstrate that both PPTases are essential for mycobac-terial viability and exhibit identical functions in mycobacteriaand corynebacteria: AcpS is responsible for the posttranslationalmodification of Fas-I, whereas PptT activates the condensingenzyme Pks13 and the various type-I Pks required for theformation of lipid virulence factors in M. tuberculosis. Theseresults allowed us to present a model for the lipid metabolism inM. tuberculosis.

ResultsMycobacteria and Corynebacteria Contain Two Conserved PPTases.Two genes encoding PPTases have been identified in the genomeof M. tuberculosis H37Rv: Rv2523c, also called acpS because theencoded 130-aa protein displays similarity to holo-ACP syn-thases (AcpS) proteins from various bacteria, and Rv2794c,which encodes a 227-aa PPTase, named PptT, involved in themycobactin biosynthesis (8, 15). Based on sequence similaritiesand conservation of amino acids motifs, PptT appeared to berelated to Sfp from Bacillus subtilis (16). We tried to identifyadditional PPTase genes by searching the M. tuberculosis H37Rvgenome with conserved PPTase motifs as probes (17). Weidentified no new gene, suggesting that M. tuberculosis containsjust two PPTases, responsible for activating at least 20 proteinsubstrates.

We then investigated whether these two PPTase genes wereconserved in other mycobacterial species and in closely related

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: DIM, phthiocerol dimycocerosates; PGL-tb, phenolic glycolipid; Pks,polyketide synthase; Fas, fatty acid synthase; P-pant, 4�-phosphopantetheine; NRPS, non-ribosomal peptide synthases; PPTase, 4�-phosphopantetheinyl transferase; ACP, acyl carrierprotein; ATc, anhydrotetracycline; Km, kanamycin.

‡To whom correspondence should be addressed. E-mail: christian.chalut@ipbs.fr.

© 2006 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0511129103 PNAS � May 30, 2006 � vol. 103 � no. 22 � 8511–8516

MIC

ROBI

OLO

GY

Dow

nloa

ded

by g

uest

on

July

23,

202

1

corynebacterial species. In all of the genomes analyzed (includ-ing eight mycobacterial species and three corynebacterial spe-cies), we found orthologs of acpS and pptT. The amino acidsequences of the encoded proteins were �80% similar for themycobacterial proteins, with �40% similarity between coryne-bacterial proteins and their M. tuberculosis counterparts. Thesebioinformatic analyses showed that the related mycobacterialand corynebacterial species contain only two highly conservedPPTases with different catalytic properties.

AcpS and PptT Are Essential for the Viability of Mycobacteriumsmegmatis. Mycobacteria contain many proteins that requireactivation to become functional: two Fas systems, at least 3NRPS and 18 type-I Pks in M. tuberculosis. Several of theseproteins are involved in biological processes essential for myco-bacterial viability. We therefore first investigated whether eitherof the PPTase genes could be disrupted in the model mycobac-terial strain M. smegmatis. Two nonreplicative plasmids carryingthe counterselectable marker sacB and a disrupted allele ofeither acpS or pptT from M. smegmatis were constructed andinserted into the chromosome by single crossover, generatingstrains PMM68 and PMM70, respectively (see Fig. 6, which ispublished as supporting information on the PNAS web site). Theplating of PMM68 or PMM70 on medium containing 5% sucroseand kanamycin (Km) did not lead to selection of a secondrecombination event, suggesting that both acpS and pptT areessential for mycobacterial viability. We produced two condi-tional M. smegmatis mutants to confirm the essential nature ofthese genes. PMM68 cells were transformed with pC-acpSms, athermosensitive mycobacterial vector carrying an intact copy ofthe acpS gene. Selection on sucrose-containing LB agar at 30°Cresulted in colonies in which the second recombination eventbetween the two chromosomal alleles had occurred. One clone,PMM77 (�acpS:pC-acpSms), containing a disrupted acpS::kmgene on the chromosome and a functional acpS gene in thethermosensitive plasmid pC-acpSms, was chosen for furtherwork (see Fig. 6). The same strategy was used to generatePMM78 (�pptT:pC-pptTms), a recombinant strain harboring amutated pptT::km chromosomal gene and an intact pptT allele inthe thermosensitive plasmid pC-pptTms (see Fig. 6).

We investigated the function in mycobacterial growth of theAcpS and PptT proteins by streaking the recombinant strainsPMM77, PMM78, and the WT strain of M. smegmatis, grown at30°C, on LB agar plates. At 30°C, PMM77 and PMM78 grew aswell as the WT. In contrast to the WT, however, neither of therecombinant strains grew at 42°C, a nonpermissive temperaturefor plasmid replication (Fig. 1). Thus, AcpS and PptT are notredundant and are clearly required for the survival of M.smegmatis, activating proteins involved in essential biosynthesispathways.

Corynebacterium glutamicum �pptT and �acpS Mutants Display Dif-ferent Microbiological Phenotypes. The synthesis of fatty acids andmycolic acids are known to be essential for the growth of M.smegmatis in laboratory conditions. We therefore hypothesizedthat the PptT and AcpS PPTases were involved in at least oneof these metabolic pathways. We investigated the roles of AcpSand PptT in fatty acid and mycolic acid biosynthesis in C.glutamicum, another model bacterium more tolerant than my-cobacterial strains to mutation in genes involved in mycolatemetabolism (9). This species has a cell envelope very similar tothat of mycobacteria but produces a mixture of saturated andunsaturated corynomycolic acids, typically containing 30 to 36carbons (3).

We investigated the roles of the two PPTases by knocking outthe acpS and pptT genes in C. glutamicum, replacing the WTalleles with the corresponding Km-disrupted genes. On BHI agarplates, the �pptT mutant produced small, rough colonies ratherthan large, shiny colonies, as observed for the WT. This mutantalso grew much slower than the WT in liquid medium, with highlevels of cell aggregation, whereas WT cells were well dispersedin the same conditions (data not shown). These phenotypicchanges were similar to those observed for the C. glutamicum�pks13 mutant, which lacks an enzyme essential for the produc-tion of corynomycolates (9) suggesting cell envelope modifica-tions consistent with an absence of mycolate production. Com-plete reversion of these phenotypic modifications was observedafter transformation of the mutant with pCGL-pptTcg, whichcontains a functional copy of pptT from C. glutamicum. Theobserved phenotype of the mutant strain was therefore clearlydue to deletion of the pptT gene.

The same strategy gave �acpS mutants only if the growthmedium was supplemented with fatty acids. Phenotypic analysisconfirmed that these C. glutamicum �acpS mutants were unableto grow on solid or liquid BHI medium unless sodium oleate andTween 40 were added. Under these conditions, no phenotypicdifferences were observed between the WT and �acpS recom-binant strains (data not shown). This auxotrophy for oleic acidsuggests that the disruption of acpS in C. glutamicum affects fattyacid biosynthesis. This phenotype was fully reversed to WT aftertransfer of pCGL-acpScg, a corynebacterial plasmid carrying afunctional acpS gene.

The AcpS and PptT PPTases Are Involved in Fatty Acid and Mycolic AcidBiosynthesis, Respectively, in C. glutamicum. C. glutamicum WT, the�acpS and �pptT mutants, and the complemented �acpS:pCGL-acpScg and �pptT:pCGL-pptTcg strains were grown to expo-nential growth phase and fatty acids were released from bacteriaby saponification of whole cells. TLC showed that the �acpSmutant synthesized corynomycolates in similar amounts to theWT and the complemented strain (Fig. 2A). The �acpS mutantwas found to contain large amounts of C16 and C18:1 fatty acids,the precursors of corynomycolates (Fig. 2B). Because thismutant was auxotrophic for fatty acids, these lipids probablyoriginated from the oleic acid and Tween 40 added to themedium. Unlike the WT (Fig. 2B) and complemented strains(data not shown), which synthesized various forms of coryno-mycolates, the �acpS mutant strain produced only one detect-able corynomycolic acid derivative (Fig. 2B) that was charac-terized by GC-MS as a C36:2 corynomycolate (see Fig. 7, whichis published as supporting information on the PNAS web site),indicating that the mutant was able to condense two C18:1 fattyacids originated from the exogenous oleic acid. The absence ofC32 and C34 corynomycolates from the cell envelope probablyresulted from the large amounts of oleate in the growth medium(18). Unlike mycobacteria, which possess two fatty acid synthasesystems (Fas-I and Fas-II), C. glutamicum has two Fas-I proteins(Fas-IA and Fas-IB) but lacks the Fas-II system. Fas-IA isessential for growth, whereas Fas-IB is not (18). The effect on

Fig. 1. Thermosensitivity of the M. smegmatis conditional mutants. M.smegmatis WT, PMM77 (�acpS:pC-acpSms), and PMM78 (�pptT:pC-pptTms)were grown in LB (with Km and streptomycin for the recombinant strains) at30°C, streaked onto LB agar plates, and incubated for 2 days at 30°C (Left) or42°C (Right).

8512 � www.pnas.org�cgi�doi�10.1073�pnas.0511129103 Chalut et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

1

fatty acid biosynthesis of disrupting acpS in C. glutamicumdemonstrates that the Fas-IA enzyme is inactive in the �acpSmutant, strongly suggesting that AcpS is responsible for the4�-phosphopantetheinylation of this enzyme. These findings areconsistent with previous results showing that AcpS specificallyactivates the two Fas-I proteins in Brevibacteruim ammoniagenes,another corynebacterial species (19). Moreover, the ability of therecombinant strain to sustain corynomycolate synthesis in thepresence of exogenous fatty acids implies that the AcpS PPTaseis not required for posttranslational modification of the con-densing enzyme Pks13.

We then investigated whether the phenotypic changes ob-served with the �pptT mutant reflected cell wall modifications.TLC and GC revealed that disruption of pptT in C. glutamicumabolished the production of corynomycolates (Fig. 2 A and B).As expected, complementation of the mutation by reintroduc-tion of a WT allele fully restored the production of all classes ofcorynomycolates (C32, C34:1, and C36:2) (data not shown). The�pptT mutant also produced large amounts of C16 and C18:1 fattyacids (Fig. 2B), indicating that PptT is not involved in posttrans-lational modification of the Fas-IA enzyme. The inability of themutant strain to synthesize mycolic acids from fatty acidssuggests that PptT is involved in activation of the condensingenzyme Pks13.

Biochemical Characterization of the 4�-Phosphopantetheinylation ofPks13 and Fas-I in C. glutamicum. Phenotypic and biochemicalanalyses of the two C. glutamicum mutants provided indirectevidence that AcpS and PptT display different functions, theformer being involved in the Fas-I activation and the latter in themodification of Pks13. We investigated the specificity of the twoPPTases by labeling the P-pant arm and visualizing its transferonto the various protein substrates. This experiment was basedon the generation in bacteria of a pool of CoA harboring a14C-radiolabeled 4�-phosphopantetheine prosthetic group. Thispool could then serve as a substrate for the two PPTases, for thespecific labeling of 4�-phosphopantetheinylated proteins.

The C. glutamicum WT strain was cultured in the presence of�-[�-14C]alanine, a precursor of CoA. Two protein bands werelabeled, with apparent molecular weights of 170 and �220 (Fig.3A Upper). The identity of the smaller protein was establishedbased on the lack of labeling of the band in a �pks13 mutant. Asearch in the genomic database of C. glutamicum ATCC13032for the presence of large proteins that must be converted by thecovalent attachment of the P-pant group revealed Fas-IA andFas-IB (molecular weight, 315 and 317) as the only candidates.We therefore concluded that the larger of the two protein bandscorresponded to the Fas-I enzymes.

When the same experiment was performed with the �pptTstrain, Pks13 was produced but not labeled, in contrast to theFas-I enzymes, which were still activated (Fig. 3A). Whenexperiments were performed with the �acpS mutant, the oppo-site labeling pattern was observed: Pks13 was labeled but Fas-Ienzymes were not (Fig. 3A Upper). These experiments estab-lished the repertoire of protein substrates for the two PPTasesof C. glutamicum. AcpS and PptT have strict substrate specificityfor the Fas-I enzymes and Pks13, respectively.

AcpS and PptT Can Perform the Same Catalytic Reactions in Myco-bacteria and Corynebacteria. The experiments described aboveestablished the function of each PPTase in corynebacterial cells.We investigated whether these proteins had similar functions inmycobacteria by testing whether the disruption of either the pptTor the acpS gene in M. smegmatis could be complemented byexpression of the corresponding ortholog from C. glutamicum.We performed this experiment by constructing two mycobacte-rial temperature-sensitive plasmids, pC-acpScg and pC-pptTcg,harboring the acpS and pptT genes of C. glutamicum, respec-tively. We introduced pC-acpScg into M. smegmatis PMM68,which is merodiploid for the acpS gene, and pC-pptTcg into M.smegmatis PMM70, which is merodiploid for the pptT gene.Several transformants were plated on solid medium containingsucrose to induce the second recombination event between thetwo chromosomal alleles at either the pptT or the acpS locus. Two

Fig. 2. Biochemical analysis of the cell envelope of C. glutamicum �acpS and �pptT recombinant strains. (A) Fatty acids and mycolic acids produced by C.glutamicum WT; the various mutants and the complemented strains were prepared from cells, derivatized into methyl esters, and separated by analytical TLCon Durasil 25TLC (Macherey & Nagel) with dichloromethane. Lipids were visualized by spraying the plates with 10% phosphomolybdic acid in ethanol andheating. (B) Trimethylsilyl derivatives of fatty acid methyl esters from C. glutamicum WT, �acpS, �pptT, and �pptT:pCGL-pptTms were analyzed by GC, asdescribed in ref. 28. For each chromatogram, the portion between 36 and 44 min is magnified. Peaks corresponding to the C16–C18:1 fatty acids and the C32, C34:1,and C36:2 corynomycolate methyl esters are indicated.

Chalut et al. PNAS � May 30, 2006 � vol. 103 � no. 22 � 8513

MIC

ROBI

OLO

GY

Dow

nloa

ded

by g

uest

on

July

23,

202

1

clones, named PMM84 (�acpS:pC-acpScg) and PMM85 (�pptT:pC-pptTcg) were selected for further analyses. PMM84 andPMM85 had a phenotype identical to that obtained with mutantsPMM77 (�acpS:pC-acpSms) and PMM78 (�pptT:pC-pptTms),which produced the AcpS and PptT proteins of M. smegmatis,respectively. Both mutant strains grew normally on LB agar platesat 30°C but neither grew at 42°C (see Fig. 8, which is published assupporting information on the PNAS web site). Thus, the PPTasesof C. glutamicum recognize and posttranslationally modify the sameessential proteins in M. smegmatis as their mycobacterial orthologs.Similar cross-complementation was observed when the M. smeg-matis PptT protein was produced in the C. glutamicum �pptTmutant. Fatty acid analysis by TLC and GC, after the saponificationof whole cells, revealed that corynomycolate production had beenrestored to WT levels in the resulting strain (Fig. 2 A and B). Thus,the M. smegmatis PptT was able to activate the Pks13 of C.glutamicum, suggesting that this protein performs the same catalyticreaction in M. smegmatis. These cross-complementation experi-ments showed that PptT and AcpS PPTases have similar functionsin the biosynthesis of fatty acids, including mycolic acids, in myco-bacteria and corynebacteria.

PptT Catalyzes the 4�-Phosphopantetheinylation of M. tuberculosisType-I Pks. We demonstrated that PptT specifically catalyzes theposttranslational modification of Pks13 in corynebacteria andmycobacteria. This protein is the only type-I Pks encoded by thecorynebacterial genome. However, the repertoire of type-I Pksis much larger in mycobacteria. These additional proteins areinvolved in the formation of extractable lipids, such as DIM orPGL-tb in M. tuberculosis, which are key virulence factors. Thisobservation led us to investigate the possible involvement ofPptT in the activation of other type-I Pks in M. tuberculosis.

Accordingly, various M. tuberculosis Pks were coproduced inEscherichia coli together with the PptT of M. tuberculosis, and weinvestigated the activation of these proteins by radiolabeling with�-[�-14C]alanine. Because preliminary experiments showed thatE. coli PPTase EntD partially activated some M. tuberculosis Pks(unpublished data), we constructed an E. coli BL21(DE3)entD-disrupted strain by allelic exchange and carried out the sameexperiments in the new strain. We focused on Pks13 initially, aprotein we already identified as a substrate of PptT. Pks13 waslabeled in E. coli cells coproducing Pks13 and PptT from M.tuberculosis but not in cells producing Pks13 alone (Fig. 3BUpper). We checked the specificity of Pks13 labeling, usingpWM35�, a derivative of pWM35 encoding a mutant of thePks13 protein in which the two serine residues (S55 and S1266)responsible for attachment of the P-pant moiety of CoA withinthe N- and C-terminal ACP domains of Pks13 have beenreplaced by alanine residues. When this mutant protein wascoproduced with PptT in E. coli BL21�entD, we observed noradiolabeling in our assay, indicating that the signal obtainedwith Pks13 resulted specifically from attachment of the P-pantmoiety of CoA to the catalytic serine residues of Pks13 (Fig. 3B).These experiments provided direct evidence that Pks13 is asubstrate of PptT in M. tuberculosis.

We then investigated the activation of other type-I Pks byPptT. Five different type-I Pks (Mas and PpsA–D) were inde-pendently coproduced with PptT in E. coli BL21�entD, and in allcases, transfer of the radiolabeled P-pant was detected (Fig. 3CUpper). The signals were weak but were reproducibly detected.The labeled proteins were identified by Western blot analysisusing an antibody raised against the poly-His tag fused to the Cterminus of each protein (see Fig. 9, which is published assupporting information on the PNAS web site). In contrast, noneof these Pks was labeled in E. coli strains lacking PptT. Thisabsence of detectable activation was not due to low levels ofprotein production because Coomassie blue staining of the gel(Fig. 3C Lower) and Western blot analysis (see Fig. 9) demon-strated that type-I Pks were produced in large amounts in theabsence of PptT.

We can therefore conclude that PptT catalyzes not only theactivation of Pks13 in M. tuberculosis but also the modificationof other type-I Pks involved in the biosynthesis of lipids requiredfor virulence. We cannot formally rule out the possibility thatthese Pks are also substrates of AcpS, but the lack of redundancyof the two PPTases suggests that this is not the case.

PptT Is Essential for the Viability of M. bovis Bacillus Calmette–Guerin.Given the proposed central role of PptT for the formation oflipids specific to the M. tuberculosis complex, we tested whetherpptT is required for survival of one of these strains. We generateda conditional �pptT knockout mutant in M. bovis bacillusCalmette–Guerin using the TetR-controlled gene expressionsystem described in ref. 20. This recombinant strain, namedPMM99 (�pptT:pC-pptTmb), contains a disrupted pptT gene onthe chromosome and a complementation plasmid harboringboth a functional pptT gene from M. bovis bacillus Calmette–Guerin under the control of the pmyc1tetO promoter and tetR,which encodes the transcriptional repressor TetR (see Fig. 10,which is published as supporting information on the PNAS website). When the recombinant strain is cultivated in the presenceof anhydrotetracycline (ATc), TetR binds ATc and pmyc1tetOmediates pptT expression. In contrast, in the absence of ATc,TetR tightly binds to the tet operators in pmyc1tetO and preventstranscription of pptT (20).

Both M. bovis bacillus Calmette–Guerin WT and the PMM99mutant strains grew well on ATc containing plates, but weobserved growth inhibition of PMM99 on plates lacking ATc incontrast to the WT (Fig. 4). Therefore, expression of pptT isrequired to sustain M. bovis bacillus Calmette–Guerin growth

Fig. 3. 4�-Phosphopantetheinylation labeling assays using �-[�-14C]alanine.Various corynebacteria or recombinant E. coli strains were cultured in thepresence of �-[�-14C]alanine and proteins from cell extracts of each strain wereseparated by SDS�PAGE. Polyacrylamide gels were stained with Coomassieblue (Lower) and placed against x-ray film for autoradiography (Upper). (A)Analysis of C. glutamicum WT, �pptT, �pks13, and �acpS. Positions of Pks13and Fas-I are indicated by arrows and arrowheads, respectively. M, 14C-methylated markers (220, 97.4, 66, 46, 30, 21.5, 14.3 kDa) (Sigma). (B–C)Analysis of strains of E. coli BL21�entD producing either Pks13 or Pks13* (S55Aand S1266A mutated Pks13) from M. tuberculosis (B) or various type-I Pks fromM. tuberculosis (C) with the WT PptT (�) from M. tuberculosis (strains trans-formed with pLSfp) or the mutated pptT-encoded PPTase (�) (strains trans-formed with pLSfp�). Positions of Pks13 and Pks13* (B) and Mas, PpsA, PpsB,PpsC, and PpsD (C) are indicated by arrows. M, 14C-methylated markers.

8514 � www.pnas.org�cgi�doi�10.1073�pnas.0511129103 Chalut et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

1

demonstrating (i) that pptT is essential for the viability of strainsof the M. tuberculosis complex and (ii) that the lack of PptTcannot be complemented by the endogenous AcpS.

DiscussionGiven the crucial role of cell envelope lipids in the biology of M.tuberculosis, tremendous efforts have been made in recent yearsto decipher the cellular processes leading to the production andtranslocation of these components. These studies have suggestedthat the unique lipids found in mycobacteria are synthesized bythe combined action of Fas systems and type-I Pks. We inves-tigated the role of two PPTases in the posttranslational modi-fication of these biosynthetic enzymes. We provide direct evi-dence that each PPTase activates a defined subset of proteinsubstrates in mycobacteria and corynebacteria and that bothenzymes are essential for mycobacterial viability. These resultsprovide insight into lipid metabolism in mycobacteria and re-lated bacteria and demonstrate the key role played by the twoPPTases in the biology of these microorganisms. Our conclu-sions are in contrast to the prediction of Sassetti et al. (21)concerning the essentiality of acpS in mycobacteria. However,their proposal was based on a statistical analysis of insertions ina transposon mutant library and recent studies have shown thatsome genes may be missed by using their method (22, 23).

We propose a model of lipid metabolism in mycobacteria inwhich AcpS is dedicated to the posttranslational modification ofFas-I and the AcpM subunit of Fas-II, whereas PptT activates thenumerous type-I Pks and NRPS of M. tuberculosis (Fig. 5). Thismodel also applies to Corynebacterium spp. except that thesebacteria contain fewer proteins requiring modification. Thismodel is supported by several lines of evidence. Our resultsclearly demonstrate that the two mycobacterial or corynebacte-rial PPTases are not redundant and have different substraterepertoires. Indeed, the two mycobacterial PPTases were foundto be independently essential for mycobacterial viability. A C.glutamicum mutant lacking acpS displayed fatty acid auxotrophyand no Fas-I activation but was able to produce mycolic acids andto activate Pks13. In contrast, pptT deletion in C. glutamicumabolished mycolic acids synthesis and Pks13 activation but hadno effect on C16–C18 fatty acids production and Fas-I modifica-tion. These data demonstrate that Pks13 is a substrate of PptTand confirm the role of AcpS in Fas-I activation, as shown in B.ammoniagenes (19). We also found that five other type-I Pksfrom M. tuberculosis, differing in size and domain structure, werealso activated by PptT, strongly supporting a model in which alltype-I Pks in M. tuberculosis are activated by PptT. Consistentwith this major role for the lipid metabolism and the nonredun-dancy of the two PPTases, we demonstrated directly that pptT isessential for the viability of a M. tuberculosis complex strain. Astudy has shown that, in vitro, PptT 4�-phosphopantetheinylates

MbtB and MbtE, two NRPS involved in mycobactin assembly(15). The activation of two other proteins encoded by themycobactin gene cluster, MbtF, a peptide synthase and MbtD, apolyketide synthase, remains speculative due to the lack ofexperimental data, but it can be inferred from our results andthose of Quadri et al. (15) that both proteins are also activatedby PptT. Finally, AcpM can be activated by the AcpS of E. coliboth in vitro and in vivo, consistent with this protein being asubstrate of AcpS in mycobacteria (24, 25). Thus, all previousresults and the results of this study are consistent with our model.

Our results also have important implications for our under-standing of the biology of mycobacteria and particularly of M.tuberculosis, which contains �20 proteins requiring 4�-phosphopantetheinylation. We defined the substrate repertoireof each PPTase and showed that both enzymes are required forthe formation of components essential for mycobacterial viabil-ity. Fas-I, which is activated by AcpS, catalyzes the synthesis ofC16–C18 fatty acids, which are incorporated into the various lipidconstituents of the plasma membrane. The synthesis of fattyacids by Fas-I is also one of the first steps of the complexbiosynthesis pathway leading to the formation of mycolates. Thispathway includes two other proteins: AcpM (a subunit of theFas-II system) and Pks13, activated by AcpS and PptT, respec-tively. Both mycobacterial PPTases are therefore required formycolate formation and, consequently, are essential for myco-bacterial viability. Both enzymes are also required for theproduction of major virulence factors. The enzymes MbtB andMbtD-F, which are activated by PptT, are involved in assemblyof the mycobactin siderophores required for growth withinhuman macrophages (26). Similarly, Fas-I and seven type-I Pks,activated by AcpS and PptT, respectively, are involved in the

Fig. 4. Effect of PptT depletion on growth of M. bovis bacillus Calmette–Guerin. M. bovis bacillus Calmette–Guerin WT and the PMM99 (�pptT:pC-pptTmb) mutant strain were grown in 7H9 containing ADC and ATc (50 ng�ml)(with Km and hygromycin for PMM99) at 37°C and streaked onto 7H11 platessupplemented with OADC with or without ATc (100 ng�ml). Plates wereincubated for 20 days at 37°C.

Fig. 5. Schematic diagram of the role played by AcpS and PptT in the fattyacid, mycolic acid, methyl-branched-containing lipid, and siderophore biosyn-thesis pathways in M. tuberculosis. Only proteins requiring 4�-phospho-pantetheinylation in these pathways are indicated. Proteins activated by PptTare boxed in rectangles and proteins modified by AcpS are boxed in ovals.p-HBA, para-hydroxybenzoic acid; SL-1, sulfolipid 1; PAT, polyacyltrehaloses.

Chalut et al. PNAS � May 30, 2006 � vol. 103 � no. 22 � 8515

MIC

ROBI

OLO

GY

Dow

nloa

ded

by g

uest

on

July

23,

202

1

formation of DIM and PGL-tb, two complex lipids produced bya very limited number of mycobacterial species and serving asmajor virulence factors in M. tuberculosis (7, 27).

In conclusion, the two PPTases characterized in this study areessential for the viability and pathogenesis of M. tuberculosis andprobably for other mycobacteria that include major pathogenssuch as Mycobacterium leprae, Mycobacterium ulcerans, andMycobacterium avium, all of which have orthologs of AcpS andPptT and produce mycolates and lipid virulence factors. The twomycobacterial PPTases should therefore be considered as verypromising targets for the development of drugs for treatingmycobacterial infections.

Materials and MethodsBacterial Strains and Growth Conditions. See Supporting Materialsand Methods, which is published as supporting information onthe PNAS web site.

Construction of Plasmids and Mutant Strains of M. smegmatis, C.glutamicum, and M. bovis Bacillus Calmette–Guerin. Construction ofplasmids and recombinant strains used in this study are describedin Supporting Materials and Methods. Their relevant character-istics are summarized in Table 1, which is published as support-ing information on the PNAS web site.

Biochemical Characterization of Fatty Acids in C. glutamicum Strains.C. glutamicum was cultured to exponential growth phase, andfatty acids were prepared from cells and separated by analyticalTLC, as described in ref. 28. For GC and GC-MS analyses,trimethylsilyl derivatives of fatty acids were obtained and ana-lyzed, as described in ref. 29.

Labeling of 4�-Phosphopantetheinylated Proteins in C. glutamicum.The C. glutamicum WT, �pptT, �pks13, and �acpS strains weregrown to exponential growth phase in 5 ml of CGXII minimalmedium (30) supplemented with either 0.05% Tween 80 (forWT, �pptT, and �pks13) or 0.03% Tween 40 and 0.03% sodiumoleate (for �acpS) at 30°C. We then added 5 �l of �-[�-14C]alanine (49 mCi�mmol; Sigma) and incubated the cells for

an additional 12 h. Bacteria were harvested by centrifugation,washed twice with PBS, and disrupted by agitation for 3 min ina Mini BeadBeater (Biospec Products, Bartlesville, OK). Cellextracts were centrifuged at 13,400 � g for 10 min at 4°C, andsupernatant proteins were subjected to SDS�PAGE. The poly-acrylamide gel was stained with Coomassie blue and dried, andthe radiolabeled proteins were detected by placing the dried gelagainst Kodak x-ray film.

Labeling of the Recombinant Mycobacterial Pks in E. coli. Expressionvectors pWM35, pWM35�, pETMas, pETA, pETB, pETC, andpETD were introduced, together with either pLSfp or pLSfp�,into E. coli BL21�entD. Each strain was grown in 5 ml of M9medium supplemented with Km and chloramphenicol at 30°C.When optical density at 600 nm reached 0.5, we added 5 �l of 1M isopropyl-�-D-thiogalactopyranoside and 5 �l of �-[�-14C]alanine to the culture and continued the incubation for anadditional 4 h at 30°C. We then centrifuged 1 ml of culture,washed the cells with 1 ml of 50 mM Tris�HCl, pH 8.0, andresuspended them in 100 �l of 50 mM Tris�HCl, pH 8.0,containing 0.1% Triton X-100. For labeling analysis, 10 �l of cellsuspension was incubated with 10 �l of 2� denaturing buffer at95°C for 5 min, and proteins were then separated by SDS�PAGE.After electrophoresis, polyacrylamide gels were either stainedwith Coomassie blue before drying and autoradiography orproteins were electrophoretically transferred to nitrocellulosemembranes (Gelman) and analyzed by Western blotting analysiswith an enhanced cheluminescence detection system (Amer-sham Pharmacia) by using an anti-poly-His antibody (1�2,000;Sigma) and an anti-mouse peroxidase conjugate (1�10,000;Sigma).

We thank Dr. S. Ehrt for providing plasmids pSE100 and pMC1m, Dr.A. Lemassu and H. Montrosier for assistance with GC and GC-MSanalyses, Ina Haneburger for help with the construction of the Cory-bacterium mutants, Dr. M. Daffe for critical comments regarding themanuscript, and F. Viala for technical help with figure preparation. Thiswork was funded by the Centre National de la Recherche Scientifiqueand European Commission Grant LSHP-CT-2005-018923.

1. Daffe, M. & Draper, P. (1998) Adv. Microb. Physiol. 39, 131–203.2. Barry, C. E., 3rd, Lee, R. E., Mdluli, K., Sampson, A. E., Schroeder, B. G.,

Slayden, R. A. & Yuan, Y. (1998) Prog. Lipid. Res. 37, 143–179.3. Daffe, M. (2005) in Handbook of Corynebacterium Glutamicum, eds. Eggeling,

L. & Bott, M. (Taylor and Francis, CRC, Boca Raton, FL), pp. 121–148.4. Banerjee, A., Dubnau, E., Quemard, A., Balasubramanian, V., Um, K. S.,

Wilson, T., Collins, D., de Lisle, G. & Jacobs, W. R., Jr. (1994) Science 263,227–230.

5. Camacho, L. R., Ensergueix, D., Perez, E., Gicquel, B. & Guilhot, C. (1999)Mol. Microbiol. 34, 257–267.

6. Minnikin, D. E., Kremer, L., Dover, L. G. & Besra, G. S. (2002) Chem. Biol.9, 545–553.

7. Reed, M. B., Domenech, P., Manca, C., Su, H., Barczak, A. K., Kreiswirth,B. N., Kaplan, G. & Barry, C. E., 3rd. (2004) Nature 431, 84–87.

8. Cole, S. T., Brosch, R., Parkhill, J., Garnier, T., Churcher, C., Harris, D.,Gordon, S. V., Eiglmeier, K., Gas, S. & Barry, C. E., 3rd, et al. (1998) Nature393, 537–544.

9. Portevin, D., de Sousa-D’Auria, C., Houssin, C., Grimaldi, C., Chami, M.,Daffe, M. & Guilhot, C. (2004) Proc. Natl. Acad. Sci. USA 101, 314–319.

10. Onwueme, K. C., Vos, C. J., Zurita, J., Ferreras, J. A. & Quadri, L. E. N. (2005)Prog. Lipid. Res. 44, 259–302.

11. Walsh, C. T., Gehring, A. M., Weinreb, P. H., Quadri, L. E. & Flugel, R. S.(1997) Curr. Opin. Chem. Biol. 1, 309–315.

12. Keating, T. A. & Walsh, C. T. (1999) Curr. Opin. Chem. Biol. 3, 598–606.13. De Voss, J. J., Rutter, K., Schroeder, B. G. & Barry, C. E., 3rd. (1999) J.

Bacteriol. 181, 4443–4451.14. Lambalot, R. H., Gehring, A. M., Flugel, R. S., Zuber, P., LaCelle, M.,

Marahiel, M. A., Reid, R., Khosla, C. & Walsh, C. T. (1996) Chem. Biol. 3,923–936.

15. Quadri, L. E., Sello, J., Keating, T. A., Weinreb, P. H. & Walsh, C. T. (1998)Chem. Biol. 5, 631–645.

16. Nakano, M. M., Corbell, N., Besson, J. & Zuber, P. (1992) Mol. Gen. Genet.232, 313–321.

17. Weissman, K. J., Hong, H., Oliynyk, M., Siskos, A. P. & Leadlay, P. F. (2004)ChemBioChem. 5, 116–125.

18. Radmacher, E., Alderwick, L. J., Besra, G. S., Brown, A. K., Gibson, K. J.,Sahm, H. & Eggeling, L. (2005) Microbiology 151, 2421–2427.

19. Stuible, H. P., Meier, S. & Schweizer, E. (1997) Eur. J. Biochem. 248, 481–487.20. Ehrt, S., Guo, X. V., Hickey, C. M., Ryou, M., Monteleone, M., Riley, L. W.

& Schnappinger, D. (2005) Nucleic Acids Res. 33, e21.21. Sassetti, C., Boyd, D. H. & Rubin, E. J. (2003) Mol. Microbiol. 48, 77–84.22. Movahedzadeh, F., Smith, D. A., Norman, R. A., Dinadayala, P., Murray-Rust,

J., Russell, D. G., Kendall, S. L., Rison, S. C., McAlister, M. S., Bancroft, G. J.,et al. (2004) Mol. Microbiol. 51, 1003–1014.

23. Rodriguez, G. M., Voskuil, M. I., Gold, B., Schoolnik, G. K. & Smith, I. (2002)Infect. Immun. 70, 3371–3381.

24. Wong, H. C., Liu, G., Zhang, Y. M., Rock, C.O. & Zheng, J. (2002) J. Biol.Chem. 277, 15874–15880.

25. Schaeffer, M. L., Agnihotri, G., Kallender, H., Brennan, P. J. & Lonsdale, J. T.(2001) Biochim. Biophys. Acta 1532, 67–78.

26. De Voss, J. J., Rutter, K., Schroeder, B. G., Su, H., Zhu, Y. & Barry, C. E., 3rd.(2000) Proc. Natl. Acad. Sci. USA 97, 1252–1257.

27. Cox, J. S., Chen, B., McNeil, M. & Jacobs, W. R., Jr. (1999) Nature 402, 79–83.28. Laval, F., Laneelle, M. A., Deon, C., Monsarrat, B. & Daffe, M. (2001) Anal.

Chem. 73, 4537–4544.29. Constant, P., Perez, E., Malaga, W., Laneelle, M.-A., Saurel, O., Daffe, M. &

Guilhot, C. (2002) J. Biol. Chem. 277, 38148–38158.30. Keilhauer, C., Eggeling, L. & Sahm, H. (1993) J. Bacteriol. 175, 5595–5603.

8516 � www.pnas.org�cgi�doi�10.1073�pnas.0511129103 Chalut et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

1