New insights into the biogenesis of the cell envelope of corynebacteria: identification and functional characterization of five new mycoloyltransferase genes in Corynebacterium glutamicum

Abstract

Mycolic acids, the major lipid constituents of Corynebacterineae, play an essential role in maintaining the integrity of the bacterial cell envelope. We have previously characterized a corynebacterial mycoloyltransferase (PS1) homologous in its N-terminal part to the three known mycobacterial mycoloyltransferases, the so-called fibronectin-binding proteins A, B and C. The genomes of Corynebacterium glutamicum (ATCC13032 and CGL2005) and Corynebacterium diphtheriae were explored for the occurrence of other putative corynebacterial mycoloyltransferase-encoding genes (cmyt). In addition to csp1 (renamed cmytA), five new cmyt genes (cmytB–F) were identified in the two strains of C. glutamicum and three cmyt genes in C. diphtheriae. In silico analysis showed that each of the putative cMyts contains the esterase domain, including the three key amino acids necessary for the catalysis. In C. glutamicum CGL2005 cmytE is a pseudogene. The four new cmyt genes were disrupted in this strain and overexpressed in the inactivated strains. Quantitative analyses of the mycolate content of all these mutants demonstrated that each of the new cMyt-defective strains, except cMytC, accumulated trehalose monocorynomycolate and exhibited a lower content of covalently bound corynomycolate than did the parent strain. For each mutant, the mycolate content was fully restored by complementation with the corresponding wild-type gene. Finally, complementation of the cmytA-inactivated mutant by the individual new cmyt genes established the existence of two classes of mycoloyltransferases in corynebacteria.

Corynebacterium, Mycolic acid, Mycoloyltransferase

1 Introduction

Corynebacterineae is a suprageneric actinomycetes group which includes corynebacteria, mycobacteria, nocardia, rhodococci and other related microorganisms. These Gram-positive bacteria are widely distributed in nature and are important in several ways; for instance, some bacterial species, e.g. Mycobacterium tuberculosis and Mycobacterium leprae, are known to cause severe infectious diseases in humans while serious economic losses are caused by corynebacterial diseases of animals [1,2]. In addition, some corynebacteria are of considerable economic benefit in the industrial production of amino acids [3]. All these bacteria share the property of having an unusual cell envelope composition and architecture that differ from those of other Gram-positive microorganisms [2,4–7]. In particular, the cell wall and the outer layer are typified by the presence of long-chain (up to C₉₀ in mycobacteria) α-alkyl, β-hydroxy fatty acids, called mycolic acids, which are believed to play a central role in the formation of a second permeability barrier functionally similar to the outer membrane in Gram-negative bacteria [8,9]. While the latter is a typical bilayer of phospholipid and lipopolysaccharide, in mycobacteria and corynebacteria it consists of a monolayer of mycoloyl residues covalently linked to the cell wall arabinogalactan and includes other lipids, e.g. trehalose monomycolate (TMM) and trehalose dimycolate (TDM), which are probably arranged to form a bilayer [4,5,7–9].

From recent advances in structural analysis and molecular biology of Corynebacterineae it appears that corynebacteria represent a very good model for comprehensive studies of the organization and biogenesis of the cell envelopes of these bacteria. Indeed, corynebacteria possess short-chain mycolates (C₂₂–C₃₆) [10,11] and the simplest cell envelope structure and composition but structurally and functionally close to those of mycobacteria and related genera [5]. This close functionality has been recently established in the case of specific enzymes involved in the biogenesis of mycolic acids, namely the mycoloyltransferases [12]. These enzymes, which catalyze the transfer of a mycoloyl residue onto trehalose, TMM and/or the cell wall arabinogalactan, are believed to be important for the physiology of members of the Corynebacterineae group. Three mycoloyltransferases, the fibronectin-binding proteins (Fbps, also named Ag85A–C), have been identified in mycobacteria [13]. The purified Fbps of M. tuberculosis have been shown to catalyze the transfer of a mycoloyl residue from one molecule of TMM to another TMM leading to the formation of TDM. To gain insight into the functions of Fbps in the whole bacterium, we have characterized three fbp-inactivated single mutants of M. tuberculosis[14,15] and have observed a partial redundancy between the active Fbp enzymes [15]. Alternatively, we have taken advantage of the reported existence in Corynebacterium glutamicum of only one gene, csp1, which encodes a protein, PS1, whose deduced N-terminal region is similar to that of Fbps [16]. Analysis of the csp1-disrupted mutant of C. glutamicum has established the expected function of the gene product and has facilitated the identification of the in vivo acceptors of the mycoloyltransferase activity of the individual Fbps by expressing the various antigens 85 in the csp1-disrupted mutant [12]. Unexpectedly, however, the mutant was found to produce to some extent both trehalose mycolates and mycoloylated cell wall arabinogalactan, indicating that additional mycoloyltransferases exist in C. glutamicum. A subsequent study has revealed the presence of other polypeptides in several corynebacterial species which reacted with antibodies directed against the PS1 protein [5]. The present study was thus undertaken in order to identify these genes and characterize their functions.

2 Materials and methods

2.1 Strains and media

The various strains and plasmids used are listed in Table 1. Strains of C. glutamicum were grown in brain-heart infusion (3.7%, BHI) with shaking (250 rpm) at 34°C. Transformation of C. glutamicum by electroporation was performed as described by Bonamy et al. [17]. Antibiotics were added to a final concentration of 25 µg ml⁻¹ for kanamycin (Km) and 15 µg ml⁻¹ for chloramphenicol (Cm). Escherichia coli TOP10 (Invitrogen) strain was grown on Luria–Bertani (LB) medium.

1

Strains and plasmids used

Strains/plasmids	Characteristics	Source or reference
E. coli
TOP10	F−mcrA Δ(mrr-hsdRMS-mcrBC) φ80lacZΔM15 ΔlacX74 recA1 deoR araD139 Δ(ara-leu)7697 galU galK rpsL (StrR) endA1 nupG	Invitrogen
C. glutamicum
CGL2005	Restrictionless derivative of ATCC21086 (previously named Brevibacterium lactofermentum 15), RifR	[26]
CGL2022	CGL2005 with cmytA disrupted, KmR	[16]
CGL2024	CGL2005 with cmytC disrupted, KmR	This work
CGL2026	CGL2005 with cmytB disrupted, KmR	This work
CGL2027	CGL2005 with cmytD disrupted, KmR	This work
CGL2028	CGL2005 with cmytF disrupted, KmR	This work
Plasmids
pCR®2.1-TOPO®	E. coli cloning vector f1 and pUC origins, lacZα, KmR, AmpR	Invitrogen
p::mytB	pCR®2.1-TOPO® carrying 408 bp of cmytB, KmR, AmpR	This work
p::mytC	pCR®2.1-TOPO® carrying 351 bp of cmytC, KmR, AmpR	This work
p::mytD	pCR®2.1-TOPO® carrying 514 bp of cmytD, KmR, AmpR	This work
p::mytF	pCR®2.1-TOPO® carrying 632 bp of cmytF, KmR, AmpR	This work
pCGL482	Shuttle vector E. coli/C. glutamicum, CmR	[20]
pCGL2300	Derivative of pCGL482 containing the cmytC gene of CGL2005 (1792 bp, 365 bp upstream of the start codon and 323 bp downstream of the stop codon), CmR	This work
pCGL2302	Derivative of pCGL482 containing the cmytB gene of CGL2005 (1964 bp, 756 bp upstream of the start codon and 215 bp downstream of the stop codon), CmR	This work
pCGL2304	Derivative of pCGL482 containing the cmytD gene of CGL2005 (1845 bp, 511 bp upstream of the start codon and 100 bp downstream of the stop codon), CmR	This work
pCGL2305	Derivative of pCGL482 containing the cmytF gene of CGL2005 (1858 bp, 350 bp upstream of the start codon and 35 bp downstream of the stop codon), CmR	This work

2.2 DNA manipulation

Recombinant DNA techniques were performed as described by Sambrook et al. [18]. Plasmid DNAs were isolated using Promega or Qiagen kits for DNA purification. C. glutamicum chromosomal DNA was extracted as described by Ausubel et al. [19]. Restriction endonucleases and DNA-modifying enzymes were obtained from Promega and used according to the manufacturer's instructions. Oligonucleotide primers were synthesized by Genosys. Polymerase chain reaction (PCR) amplifications were performed in a GenAmp PCR 2400 thermocycler (Perkin Elmer) using Taq (Promega) or Pfu (Promega) DNA polymerases and for fragments of more than 3 kb long, JumpStart REDAccuTaq™ (Sigma) DNA polymerase. PCR fragments were systematically purified (Qiaquick PCR purification kit, Qiagen). All DNA sequencing was carried out by Q-biogene and Genome Express.

2.3 Identification of putative cmyt genes in C. glutamicum CGL2005 strain

PCRs were performed using six pairs of degenerated oligonucleotides (designed from multiple alignments of cmyt genes and protein sequences of Corynebacterium diphtheriae and M. tuberculosis) with CGL2005 genomic DNA as a matrix. The resulting products were cloned into plasmid pCR®2.1-TOPO® using the TOPO TA cloning® kit (Invitrogen). Clones carrying inserts of the expected size (between 500 and 800 bp from the known sequences of C. diphtheriae) were sequenced. To obtain the sequences of the entire genes and their flanking regions, amplifications by inverse PCR were performed using primers which were designed pointing outward from the internal cmyt sequences obtained by degenerate PCR. As a template, chromosomal DNA from C. glutamicum CGL2005 was completely digested with PstI or SalI or BglII and then circularized with T4 DNA ligase. Depending of their purity the inverse PCR products were sequenced directly or after cloning into plasmid pCR®2.1-TOPO® and sequenced.

2.4 Disruption of cmyt genes

To disrupt the cmyt genes, internal fragments of cmytB, cmytC, cmytD and cmytF were amplified from C. glutamicum CGL2005 chromosomal DNA and cloned into plasmid pCR®2.1-TOPO®. The four resulting plasmids (p::cmytB, p::cmytC, p::cmytD and p::cmytF) were used to electrotransform strain CGL2005 of C. glutamicum. Transformants generated by integration of the plasmid were selected on BHI plates containing 25 µg ml⁻¹ Km. Genomic DNAs from mutants were analyzed by three PCR amplifications using two primers localized upstream of the 5′-end and downstream of the 3′-end of the cmyt gene and the M13 forward and reverse primers localized in the pCR®2.1-TOPO® vector. The PCR products corresponding to the gene disruption borders were systematically sequenced.

2.5 Construction of plasmids carrying cmyt genes

Each of the cmyt genes was amplified from C. glutamicum CGL2005 genomic DNA. The amplified regions are as follow: 2977 bp for cmytA with BamHI and XhoI artificial sites at each extremity of the fragment, 1964 bp for cmytB, 1792 bp for cmytC with PstI artificial sites at each extremity of the fragment, 1845 bp for cmytD, 1858 bp for cmytF. PCR fragments containing the cmytA and cmytC genes were digested with BamHI/XhoI and PstI, respectively, and inserted into pCGL482 [20] digested with the same restriction enzymes to give pCGL2303 (cmytA) and pCGL2300 (cmytC). PCR fragments containing the cmytB, D and F genes were subcloned into plasmid pCR®2.1-TOPO® and fragments were re-isolated using TOPO vector restriction sites, i.e. BamHI/XhoI for cmytB, HindIII, whose end was made blunt with Klenow fragment, and PstI for cmytD and EcoRI, whose end was made blunt with Klenow fragment, and BglII for cmytF. The resulting inserts were purified and ligated to pCGL482 digested with the appropriate enzymes (SmaI in the case of blunt ends) to give pCGL2302 (cmytB), pCGL2304 (cmytD) and pCGL2305 (cmytF).

2.6 Isolation, fractionation and analysis of whole cell lipids

Lipids were obtained and analyzed as previously described [12]. Briefly, lipids were extracted from wet cells for 16 h with CHCl₃/CH₃OH (1:1, v/v) at room temperature with continuous stirring; the bacterial residues were re-extracted three times with CHCl₃/CH₃OH (2:1, v/v) and the organic phases were pooled and concentrated. The crude lipid extracts were partitioned between the aqueous and the organic phases arising from a mixture of CHCl₃/CH₃OH/H₂O (8:4:2, v/v); the lower organic phases were collected, evaporated to dryness to yield the crude lipid extracts from each strain and comparatively examined by thin-layer chromatography (TLC) on silica gel-coated plates (G-60, 0.25 mm thickness, Merck) developed with CHCl₃/CH₃OH (9:1, v/v) or CHCl₃/CH₃OH/H₂O (30:8:1 or 65:25:4, v/v). Detection of all classes of lipids was done by spraying the TLC plates with either rhodamine B or 20% H₂SO₄ in water, the latter followed by heating at 110°C; glycolipids were revealed by spraying plates with 0.2% anthrone (w/v) in concentrated H₂SO₄, followed by heating at 110°C. The Dittmer–Lester reagent was used for visualizing phosphorus-containing lipids.

The various classes of extractable lipids were also analyzed by labelling; briefly, 5 µCi of [¹⁴C]acetate (2.11 GBq mmol⁻¹, Amersham, France) were added to 100 ml exponential phase-grown bacteria for 1, 2 or 5 h. The reaction was stopped by centrifugation (10 min at 8000×g) and the cell pellets were extracted with CHCl₃/CH₃OH (9:1, v/v) for 16 h. The organic solutions were separated from the delipidated cells by filtration and dried; the crude lipid extracts were resuspended in CHCl₃ and analyzed by TLC using CHCl₃/CH₃OH/H₂O (30:8:1, v/v) as the developing solvent. Radioactivity was located and counted on plates using an automatic TLC linear analyzer (Berthold LB 2832). Then, the lipids were visualized by spraying with anthrone followed by charring.

Quantification of corynomycolic acids was performed as described [12]; briefly, delipidated cells (1.5 g dry weight) and lipids extracts (100 mg) of the various strains were dried under vacuum prior to weighing and saponified. The saponified products were acidified with 20% H₂SO₄ and the resulting fatty acids were extracted with diethyl ether, converted to methyl esters with diazomethane and dried under vacuum, dissolved in petroleum ether and applied to a Florisil (60–100 mesh, Merck) column equilibrated in petroleum ether. The column was irrigated stepwise with increasing concentrations of diethyl ether in petroleum ether. Fractionations were monitored by TLC on silica gel-coated plates using dichloromethane and fractions containing the same lipid compounds (non-hydroxylated fatty acid methyl esters or corynomycolates) were pooled and weighed. Three sequential determinations from separate preparations of delipidated cells were performed [12].

3 Results

3.1 Tracking the mycoloyltansferase-encoding genes in the genomes of C. diphtheriae and C. glutamicum and in silico analysis

To date the only characterized mycoloyltransferase in Corynebacterium was the PS1 protein from strain ATCC17965 of C. glutamicum, for which the corresponding csp1 gene was originally cloned and sequenced [16]. At the beginning of this work the only corynebacterial genomic information available was the genome of C. diphtheriae NCTC13129 at the Sanger center (http://www.sanger.ac.uk). In order to identify genes encoding mycoloyltransferases in this bacterium, BLAST searches were performed against this genome using PS1 as a query sequence [21]. Four DNA regions were found to give significant homology with PS1 when translated. Prediction of the open reading frames (ORFs) showed that one of the putative protein was a PS1 orthologue. The three others were shorter and homologous to the N-terminal part of PS1 and to the mycobacterial Fbps (Table 2, Fig. 1). In order to homogenize the nomenclature of this protein family in Corynebacterium we propose to name these proteins ‘cMyt’ for corynebacterial mycoloyltransferases. Accordingly, PS1 was renamed cMytA and the three new proteins were called cMytB, cMytC and cMytD.

2

Occurrence of putative mycoloyltransferase genes (cmyts) and proteins (cMyts) in different strains of Corynebacterium

Gene	Presence of cmyt genes and number of amino acids of the precursor protein in each species
	C. glutamicum ATCC13032	C. glutamicum CGL2005	C. diphtheriae	Other strains of Corynebacterium
cmytA	Present	Present	Present	C. glutamicum ATCC17965
	657 aa	657 aa	638 aa	Gene Cop1 (formerly csp1), prot PS1
	NP_602067			657 aa
				Q01377
cmytB	Present	Present	Present	No other sequence data
	341 aa	340 aa	355 aa
	NP_602069
cmytC	Present	Present	Present	C. ammoniagenes ATCC6872
	365 aa	368 aa	355 aa	Gene cspA, cell surface protein A
	NP_599594			358 aa
				BAB62413
cmytD	Present	Present	Present	No other sequence data
	411 aa	411 aa	406 aa
	NP_600254
cmytE	Present	Present (pseudo)	Absent	No other sequence data
	483 aa	stop in the protein after aa 180
	NP_600150
cmytF	Present	Present	Absent	No other sequence data
	483 aa	492 aa
	NP_601385

When available the accession number of proteins in data bases is given. For C. diphtheriae cMyt ORFs were identified as described in the text from the chromosome sequencing data of strain NCTC13129 at the Sanger center.

1

Multiple alignment of the mycoloyltransferase (Myt) protein family. cMyt sequences are from ATCC13032 (references listed in Table 2) and FbpC from M. tuberculosis H37Rv (accession number at SwissProt: P31953). Identical amino acids are in dark gray whereas similar amino acids are in light gray. The ends of the sequences are not shown because no more identity or similarity could be detected. The stars indicate the three amino acids involved in the catalytic mechanism. The underlined sequences represent the putative signal peptides.

Amino acid alignments of all these sequences revealed relatively conserved blocks from which degenerated primers were designed. The combined use of degenerate oligonucleotide-primed PCR and inverse PCR allowed us to identify and sequence three out of the four cmyt genes of C. glutamicum strain CGL2005. The ORF predictions and their translation indicated that these genes encode proteins which correspond respectively to cMytA, cMytB and cMytC (Table 2). We were unable, however, to identify the equivalent of cMytD by this PCR-based method. During the course of the study, the genome of C. glutamicum (strain ATCC13032) was published at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov). BLAST analysis of this genome yielded two additional putative mycoloyltransferases which were called cMytE and cMytF. Sequence comparisons at the nucleotide level showed 94% identity between the coding regions of cmytE and cmytF, indicating a recent duplication event and leading to two nearly identical proteins. Like cMytB–D, cMytE and cMytF are shorter than cMytA and homologous to FbpA–C, and to the N-terminal part of cMytA. From the analysis of gene locations on the genome of strain ATCC13032, it appeared that, except for cmytA and cmytB, which are separated by one ORF transcribed in a direction opposite to that of the two genes, the other cmyt genes are distributed all over the chromosome. PCR amplifications performed on the genomic DNA of C. glutamicum CGL2005, with primers homologous to the published DNA sequences of C. glutamicum ATCC13032, showed that cmytD, cmytE and cmytF were also present in C. glutamicum CGL2005. Sequencing of the corresponding genes revealed that cmytE has an in-frame stop codon at position 516 from the predicted initiation codon but otherwise is apparently intact, suggesting that this is a relatively recent mutation. It follows that cmytE is a pseudogene in C. glutamicum CGL2005.

The in silico analysis of the cMyt sequences indicated that all the proteins described in Table 2 belonged to the esterase family (PFAM accession number: PF00756[22]) and contained the three conserved catalytic triad residues (Ser, His, Glu, indicated by stars in Fig. 1) which have been shown to be critical for the enzyme activity of the mycobacterial FbpC [13], suggesting that four out of the five new cMyts identified would be functional proteins; all the cMyt proteins also possess a putative peptide signal [23], as expected from their similarity with PS1, a protein that has been demonstrated to be exported and matured during translocation [16]. The presence of peptide signal in these proteins is also consistent with their potential activities that are presumably localized in the outermost compartment of the bacterial cells.

3.2 Disruption of cmyt reading frames in C. glutamicum CGL2005

To test whether the postulated cmyt genes encode proteins with mycoloyltransferase functions, the genes were disrupted in C. glutamicum. The mutants arose from a single homologous recombination event between the chromosomal wild-type copies of the target genes and internal segments of these genes, i.e. gene segments truncated on both sides (Fig. 2A). In order to disrupt the gene upstream of the catalytic triad, the fragments were all chosen at the beginning of the coding phase. The resulting plasmids (p::cmytB–F, Table 1), able to replicate in E. coli but not in C. glutamicum, were used to electrotransform strain CGL2005 of C. glutamicum and Km^R transformants were selected. The proper integration of the plasmid was evaluated by PCR using two primers localized on the p::cmyt and flanking the insertion sites and two primers localized outside the cMyt ORF (Fig. 2B). At least one clone for each Δcmyt mutant yielded products of the expected size when amplifications were realized with Mup-Tp1, Mdo-Tp2 and Mup-Mdo (data not shown). In any case a product corresponding to the intact cmyt gene could be obtained with Mup-Mdo. In all cases the sequencing of the Mup-Tp1 and Mdo-Tp2 showed the correct integration of the plasmid into the cmyt gene. These results confirmed that all the cmyt genes were properly disrupted.

2

Disruption of the cmyt genes. A: A schematic representation of one cmyt gene from the wild-type C. glutamicum strain CGL2005. The mutants arose from a single homologous recombination event between the chromosomal wild-type copies of the target genes and internal segments of these genes, i.e. gene segments truncated on both sides. The gray box represents the amplified region which was cloned in the pCR®2.1-TOPO® vector to give the corresponding suicide vector p::cmyt. B: The disrupted cmyt gene obtained after insertion of p::cmyt by homologous recombination. The four primers used for the verification of the correct integration of p::cmyt are indicated. Tp1 and Tp2 correspond to M13 forward and reverse primers or conversely depending on the orientation of the cloned cmyt PCR fragments into the pCR®2.1-TOPO®.

3.3 Phenotypic analyses of the mutants

The various mutants showed no phenotypic difference when compared to their parent CGL2005 strain and to the cmytA-inactivated CGL2022 mutant (data not shown). All the strains shared the same growth rate and cell sizes as visualized by light microscopy. They also exhibited the same ultrastructural appearance by transmission electron microscopy [5], and freeze-fractured electron microscopy [24]. The potential effects of mutations in cmyts on the cell envelope architecture were investigated by the analysis of PS2, the major corynebacterial secreted protein that forms the crystalline surface layer in several corynebacterial species and is anchored in the cell wall, presumably in the external mycolic acid bilayer, via a C-terminal hydrophobic sequence of 21 residues [20,24]. The various isogenic cmyt-disrupted single mutants of C. glutamicum were all found to exhibit the typical ordered arrays produced by the S-layer protein PS2 at their surface as did their parent strain CGL2005 [20,24]. Finally, sodium dodecyl sulfate–polyacrylamide gel electrophoresis analysis of cell wall extracts showed no significant difference between the parent and the single mutant strains (data not shown).

3.4 Analysis of the extractable lipids of the various cmyt-inactivated mutants

To measure the effects of the disruption of the new putative cmyt genes on the bacterial physiology, the various mutants were biochemically analyzed and compared to their parent CGL2005 strain and the previously characterized cmytA-inactivated CGL2022 strain [12]. To determine the impact of the mutations on the trehalose lipids, the major mycolate-containing substances in corynebacteria, cells from the exponential growth phase were harvested at the same optical density and their organic solvent-extractable lipid contents were determined (Table 3). The native extractable lipids were analyzed by TLC and exhibited qualitatively similar profiles; these consisted of neutral lipids, trehalose monocorynomycolate (TMCM), trehalose dicorynomycolate (TDCM) and phospholipids. The cmytA-inactivated strain obviously accumulated more TMCM than did the other strains (Fig. 3). Quantification of trehalose lipids was performed by radiolabelling with acetate to determine the balance between the metabolically related TMCM and TDCM. All the cmyt-inactivated mutant strains, except the cmytC-disrupted mutant strain, elaborated relatively less TDCM (and more TMCM) than did the parent strain (Table 3). Consistent with the data from TLC analysis, the cmytA-disrupted mutant was the most affected in the synthesis of the precursor of TDCM (four-fold increase of TMCM), indicating that cMytA was the main protein involved in the synthesis of TDCM by transferring a corynomycoloyl residue onto TMCM; however, the other cMyts also contributed to the production of TDCM (Table 3).

3

Lipid analysis of wild-type, cmyt-inactivated mutants and complemented strains of C. glutamicum

Strain	Relevant phenotype	Lipid extract	TDCM/TMCM	Wall-linked mycolates
CGL2005	Wild-type	7.0±0.1	0.71	100±0
CGL2024	cmytC-disrupted	8.5±0.3	0.68	108±2
CGL2024(pCGL2300)	cmytC-disrupted, complementation with cmytC	8.4±0.2	0.75	102±3
CGL2026	cmytB-disrupted	6.9±0.3	0.30	53±4
CGL2026(pCGL2302)	cmytB-disrupted, complementation with cmytB	7.2±0.4	1.2	105±3
CGL2027	cmytD-disrupted	6.4±0.5	0.34	59±3
CGL2027(pCGL2304)	cmytD-disrupted, complementation with cmytD	9.9±1.0	0.68	91±7
CGL2028	cmytF-disrupted	6.1±0.7	0.43	68±5
CGL2028(pCGL2305)	cmytF-disrupted, complementation with cmytF	8.4±1.0	0.84	103±3

Percentages of the dry mass of extractable lipids relative to that of cells. The values correspond to the means of at least three independent determinations.

Cells were labelled with [¹⁴C]acetate and the radioactivity incorporated in trehalose dicorynomycolate (TDCM) and trehalose monocorynomycolate (TMCM) was determined at various time points. Data from a representative experiment (5 h labelling) are shown; the molar ratio of TDCM was obtained by dividing the radioactivity in TDCM by a factor 2, based on the occurrence of two molecules of mycolates in these molecules while only one mycoloyl residue is found in TMCM.

Corynomycolates were purified by column chromatography and weighed as previously described [15]. The corynomycolate content of delipidated cells (data from at least three independent determinations) is expressed relative to that of the parental CGL2005 (100%).

3

TLC of trehalose lipids from cmyt-inactivated single mutants of C. glutamicum. Exponential phase-grown parental (CGL2005) and isogenic mutant strains were extracted three times with CHCl₃/CH₃OH and the combined lipid extracts were washed, dried and weighed. Comparable amounts of lipids (100 µg) were dissolved in CHCl₃ and loaded on TLC plates (silica gel-coated plates; G-60, 0.25 mm thickness, Merck) and run in CHCl₃/CH₃OH/H₂O (65/25/4, v/v). Glycolipid spots were visualized by spraying anthrone, followed by charring. PL, phospholipids. Lanes 1–6: CGL2005 (wild-type), CGL2022 (cmytA-disrupted), CGL2024 (cmytC-disrupted), CGL2026 (cmytB-disrupted), CGL2027 (cmytD-disrupted), CGL2028 (cmytF-disrupted), respectively.

Complementation of the different mutants with the corresponding wild-type cmyt gene fully restored the original balance between TDCM and TMCM (Table 3). Furthermore, the overproduction of the cMytB, cMytD or cMytF proteins in the cmytA-disrupted mutant strain also completely restored the original balance between TDCM and TMCM (Fig. 4), proving the functional redundancy of the active cMyt proteins in terms of synthesis of trehalose mycolates. The lack of effect of the disruption of the cmytC gene on the trehalose mycolates could be due either to a lack of transcription of the gene or, alternatively, to a redundancy of this gene, as recently demonstrated for FbpA and FbpB [15]. To test these hypotheses, the cmytC gene was cloned in a replicative vector driven by its own promoter and expressed either in the cmytC-disrupted CGL2024 mutant, the cmytA-inactivated CGL2022 strain or the wild-type CGL2005 strain. While the overproduction of cMytC was obvious in CGL2024, CGL2005 and CGL2022 (data not shown), no change was observed in the TDCM/TMCM ratio of the transformants (Table 3 and Fig. 4). It was thus concluded that cMytC is virtually not involved in the transfer of mycoloyl residues onto TMCM to yield TDCM.

4

Lipid analysis of the cmytA-inactivated (CGL2022) mutant strain of C. glutamicum and its derivatives. The wild-type (CGL2005) strain, cmytA-inactivated (CGL2022) mutant strain, CGL2022 overexpressing cMytC (pCGL2300), cMytB (pCGL2302), cMytD (pCGL2304) and cMytF (pCGL2305) were grown, labelled and their lipids were extracted with organic solvents. The percentages of total extractable lipids (hatched bars) and of wall-linked corynomycolate (dotted bars) were determined relative to those of the parental strain CGL2005 (100%). Filled bars represent the molar ratios of TDCM versus TMCM that were calculated as described in Table 3.

3.5 Cell wall-linked mycolate contents of the various cmyt-disrupted mutants

We have previously shown that cMytA transfers mycoloyl residues on both TMCM and cell wall arabinogalactan. To evaluate the contribution of the individual cMyt on the transfer of mycoloyl residues onto the cell wall arabinogalactan, extensively delipidated cell residues from the various strains, which consisted primarily of cell walls, were saponified; the resulting corynomycolic acids were extracted and esterified with diazomethane. Corynomycolates were purified by chromatography on Florisil columns and the cell wall-linked mycolate contents of the strains were determined by weighing. All the mutant strains, except the cmytC-disrupted mutant, exhibited lower contents of cell wall-linked mycolates, compared to their parent CGL2005 strain (Table 3). The decrease in the cell wall-linked mycolate contents was reversed by complementation of each of the mutant strains with the corresponding wild-type cmyt gene. The functionality of the various cMyts was further studied by the overproduction of cMyt proteins in the cmytA-inactivated CGL2022 mutant. Surprisingly, however, analysis of the resulting transformants showed that while the overexpression of cmytB compensated the defect in covalently linked mycolates of the CGL2022 strain, the expression of either cMytC, cMytD or cMytF in CGL2022 did not restore the phenotype of the parent CGL2005 strain (Fig. 4). Together these data indicate that, although cMytA, cMytB, cMytD and cMytF are all involved in the transfer of corynomycoloyl residues onto the cell wall arabinogalactan, the active cMyt proteins are only partly redundant for this function. Again, the disruption of the cmytC gene or its expression in both the cmytA-inactivated CGL2022 and the cmytC-inactivated strains had no effect on the transfer of mycoloyl residues onto the cell wall arabinogalactan (Table 3, Fig. 4). It was thus concluded that cMytC was neither involved in the transfer of corynomycoloyl residues onto the cell wall arabinogalactan nor redundant with cMytA.

4 Discussion

The present study was undertaken in order to seek for and identify putative corynebacterial mycoloyltransferase (cmyt) genes other than the csp1 gene. A deliberate search for csp1-like genes in several corynebacterial species by conventional molecular techniques, combined with bioinformatic analysis of the available genome sequences, led to the identification of a total of six putative mycoloyltransferase genes in two strains of C. glutamicum, four of which are also present in C. diphtheriae. It should be noticed that the occurrence of six myt genes in corynebacteria, compared to four fbp genes in mycobacteria, was unexpected since the cell envelope of corynebacteria is much less complex than that of mycobacteria [5] and contains simple versions of the various lipids typifying the Mycobacterium genus, including mycolic acids [10,11]. The various cmyt genes were present on the genomes of the two strains of C. glutamicum examined in only one copy, with the possible exception of cmytE whose sequence was almost identical to that of cmytF. Nevertheless in C. glutamicum CGL2005 cmytE is a pseudogene. The singularity of cmytA, which encodes the cMytA protein that was present in all the corynebacterial strains examined and was much larger than cMytB–F, certainly explains why the other cmyt genes have not been detected by Joliff et al. [16]. The in silico analysis of the cMyt sequences indicated that all the proteins belonged to the esterase family and contained the three conserved catalytic triad residues (Ser, His, Glu) which have been shown to be critical for the mycoloyltransferase activity of FbpC and cMytA [12,13], suggesting that cMytB–D and cMytF might be functional proteins.

To determine the functionality of the new cMyt proteins the various coding genes were inactivated and quantitative analyses of the non-covalently linked lipids and cell wall-bound corynomycolates were done on the resulting cmyt-disrupted mutants. Our results showed that, except for cMytC, all the cMyts were active proteins and contributed to the transfer of mycoloyl residues both onto trehalose monocorynomycolate and onto cell wall arabinogalactan. The corresponding cmyt-inactivated mutants, except for the cmytC-disrupted strain, elaborated more trehalose monocorynomycolate and less trehalose dicorynomycolate and exhibited a significant decrease in the cell wall-linked mycolates. These effects were fully reversed by complementation with the wild-type gene of the corresponding cmyt gene.

In sharp contrast with what was observed with trehalose mycolates, the different cmyt genes are not functionally equal for transferring mycoloyl residues onto the cell wall acceptor. For instance, only the overexpression of cmytB, but not cmytC, cmytD and cmytF, in the cMytA-defective mutant could restore the normal wall-linked mycolates (Fig. 4). Thus, two classes of cMyts exist in corynebacteria: (i) cMytA and cMytB which can complement each other and compensate for the absence of any of the four proteins and (ii) cMytD and cMytF which can replace the other cMyt proteins for the production of trehalose mycolates but are unable to complement the cell wall defect caused by the absence of cMytA. Among the possible explanations for the occurrence of two classes of cMyts, an attractive hypothesis consists in a structural difference between the two groups that would lead to a poor affinity of cMytD and cMytF for the cell wall arabinogalactan but not for trehalose monomycolate. Alternatively, it is possible that both cMytD and cMytF may be able to transfer mycoloyl residues onto the cell wall arabinogalactan only after mycoloyl residues have been transferred to some specific positions of non-reducing arabinosyl termini by cMytA and/or cMytB. The full determination of the structural motifs that composed the corynebacterial arabinogalactan is a prerequisite step for testing the latter hypothesis. Besides, the well-expressed cmytC gene product has no effect on the transfer of mycoloyl residues as revealed by the analysis of the overproduction of cMytC in both the wild-type and cmytA-disrupted mutant strains. It remains, however, that the expressed cMytC protein, for yet unknown reasons, may adopt a non-active conformation, despite the conservation of the three key amino acids involved in the catalysis. Finally, growth measurements showed that inactivation of one single functionally active cmyt gene did not result in the inhibition of the bacterial growth, as observed in mycobacteria [14,15,25]. However, a trehalose analog, which inhibits up to 60% of the mycoloyltransferase activity of FbpC in vitro and to a certain extent the synthesis of mycolates in vivo, has been shown to have a bacteriostatic action on Mycobacterium aurum[13]. It is thus tempting to speculate that the simultaneous disruption of more than one cmyt and fbp gene would lead to a profound effect on the cell wall structure of Corynebacterineae. Ongoing experiments should clarify this point.

Acknowledgements

We are grateful to Drs. Mohamed Chami (University of Basel, Switzerland) and Pierre Gounon (INSERM U 452, Nice, France) for electron microscopy analyses. This work was supported by the Centre National de la Recherche Scientifique (CNRS, France) and the European Economic Community (Grant TB Vaccine Cluster QLK2-CT-1999-01093).