Identification of apolipoprotein n-acyltransferase (LNT) in mycobacteria

Lipoproteins of Gram-negative and Gram-positive bacteria carry a thioether bound diacylglyceryl but differ by a fatty acid amide-bound to the alpha-amino group of the universally conserved cysteine. In Escherichia coli the N-terminal acylation is catalyzed by the N-acyltransferase Lnt. Using E. coli Lnt as a query in a BLASTp search, we identified putative lnt genes also in Gram-positive mycobacteria. The Mycobacterium tuberculosis lipoprotein LppX, heterologously expressed in Mycobacterium smegmatis, was N-acylated at the N-terminal cysteine, whereas LppX expressed in a M. smegmatis lnt::aph knock-out mutant was accessible for N-terminal sequencing. Western blot analyses of a truncated and tagged form of LppX indicated a smaller size of about 0.3 kDa in the lnt::aph mutant compared to the parental strain. MALDI-TOF/TOF analyses of a trypsin digest of LppX proved the presence of the diacylglyceryl modification in both strains, parental strain and lnt::aph mutant. N-acylation was found exclusively in the M. smegmatis parental strain. Complementation of the lnt::aph mutant with M. tuberculosis ppm1 restored N-acylation. The substrate for N-acylation is a C16 fatty acid while the two fatty acids of the diacylglyceryl residue were identified as C16 and C19:0 fatty acid, the latter most likely tuberculostearic acid. We demonstrate that mycobacterial lipoproteins are triacylated. For the first time to our knowledge, we identify Lnt activity in Gram-positive bacteria and assigned the responsible genes. In M. smegmatis and M. tuberculosis the open reading frames are annotated as MSMEG_3860 and M. tuberculosis ppm1, respectively. Lipoproteins of Gram-negative and Gram-positive bacteria carry a thioether bound diacylglyceryl but differ by a fatty acid amide-bound to the α -amino group of the universally conserved cysteine. In Escherichia coli the N-terminal acylation is catalyzed by the N acyltransferase Lnt. Using E. coli Lnt as a query in a BLASTp search, we identified putative lnt genes also in Gram-positive mycobacteria. The Mycobacterium tuberculosis lipoprotein LppX, heterologously expressed in Mycobacterium smegmatis , was N -acylated at the N-terminal cysteine, whereas LppX expressed in a M. smegmatis lnt::aph knock-out mutant was accessible for N-terminal sequencing. Western blot analyses of a truncated and tagged form of LppX indicated a smaller size of about 0.3 kDa in the lnt::aph mutant compared to the parental strain. MALDI-TOF/TOF


INTRODUCTION
Proteins of various organisms are modified in numerous ways, one of them is lipidation. Lipid modification of proteins is common in eucaryal and bacterial organisms and can involve myristoyl, palmitoyl, isoprenyl polymers of various lengths or aminoglycan-linked phospholipids (1,2). Lipoprotein modifications investigated here are restricted to bacteria. The lipoprotein biosynthesis pathway is a major virulence factor in Mycobacterium tuberculosis, the causative agent of human tuberculosis. Every year 1.6 million people fall prey to tuberculosis and one third of the world's population is infected (http://www.who.int/mediacentre/factsh eets/fs104/en/index.html). Thus, tuberculosis is responsible for 2.5 % of deaths in the world, which is the highest rate claimed by a single infectious agent. An M. tuberculosis knock-out mutant deficient in lipoprotein signal peptidase lspA showed reduced multiplication in bonemarrow derived macrophages, complete absence of lung pathology and a 1000 fold reduced number of colony forming units in a mouse model of infection (3,4). Likewise, lipoprotein synthesis contributes to virulence of other Gram-positive pathogens, Listeria, Staphylococci and Streptococci (5). Bacterial lipoproteins are a functionally diverse class of lipidated proteins involved in cell wall synthesis, nutrient uptake, adhesion and transmembrane signalling (6) and about 2 % of open reading frames encode this kind of proteins (7). Lipidation allows anchoring of these proteins to the cell surface. Lipoproteins are characterized by the presence of a consensus sequence, the "lipobox", located in the C-terminal part of the leader sequence and consisting of four amino acids [LVI/ASTVI/GAS/C] (7). Precursor lipoproteins are mainly translocated in a Sec-dependent manner across the plasma membrane and modified subsequently on the universally conserved, essential cysteine residue located in the lipobox motif. The modifications taking place after translocation are consecutively mediated by three enzymes: 1. formation of a thioether linkage between the conserved cysteine residue and a diacylglycerol catalyzed by phosphatidylglycerol: pre-prolipoprotein diacylglyceryl transferase (Lgt), 2. cleaveage of the N-terminal signal peptide by prolipoprotein signal peptidase/signal peptidase II (LspA) and 3. in case of Gram-negative bacteria, aminoacylation of the N-terminal cysteine residue by phospholipid:apolipoprotein N-acyltransferase (Lnt) (6)(7)(8). In E. coli, most of the mature triacylated lipoproteins are finally transported across the periplasm by the LolABCDE transport system (9). Homologues of the Lol-transport system are absent in Mycobacteria. Although lipoprotein modifying enzymes act sequentially, Lgt-independent LspA-mediated signal sequence cleveage has recently been demonstrated in Listeria monocytogenes (10). While Lgt and LspA are universally present in both, Gram-positive and Gram-negative bacteria, Lnt has been reported to be restricted to Gram-negatives (11), although some indications for N-acylation in Bacillus subtilis and Staphylococcus aureus were reported (12-15). Mycobacterial lipoproteins are immunodominant antigens (16) and several manipulate innate immune mechanisms and antigen presenting cells (17). It is known that mycobacterial lipoproteins, e.g. the 19 kDa lipoprotein, activate toll-like receptor 2 (TLR2) and co-receptors TLR1, which recognize triacylated peptides, but also TLR6, which recognize diacylated peptides (18,19). However, the lipid linkage of mycobacterial lipoproteins has not been determined. In this study, we show that Lnt activity is more widely distributed than previously assumed. We demonstrate apolipoprotein N-acyltransferase activity in a Gram-positive Mycobacterium and give complete structural information about the lipid modification of mycobacterial lipoproteins. Hereby, the functionality of Lnt homologues in Actinomycetes is revealed (5). We show that mycobacterial lipoproteins are triacylated and carry mycobacteria specific fatty acids.
Complementation of conditional E. coli lnt mutant PAP8508 LntMs was amplified by PCR and cloned into the EcoRI / BamHI sites of pUC18 resulting in pUC18-lntMs. Plasmids pUC18-lntMs323W, pUC18-lntMs477Y and pUC18-lntMs323W/ 477Y were generated by standard mutagenesis-PCR techniques. The E. coli conditional lnt mutant PAP8508 and its parental strain PAP105 (a generous gift of N. Buddelmeijer) were used for complementation analysis (22). Strains were plated on LB agar supplemented with 1 mM IPTG, 100 μg / ml ampicillin and either 0.4 % (w/v) glucose or 0.2 % (w/v) arabinose.  (20) to result in ptrpA1-rpsL-lntMs::aph. The lntMs::aph allele was substituted for lntMs in the M. smegmatis chromosome as described previously (23) and confirmed by Southern blot analyses with a 0.2 kbp SmaI / NcoI lntMs upstream probe. For complementation with M. smegmatis lnt, a 4.3 kbp PvuII fragment from pGem-T Easy-lntMs comprising the entire lntMs gene was cloned into the HpaI site of plasmid pMV361-hyg (24) to result in pMV361-hyg-lntMs. For complementation with M. tuberculosis ppm1 a 6.3 kbp fragment from M. tuberculosis' genomic position 2'306'187 to 2'312'526 spanning the entire ppm1 gene was cloned into pGem-T Easy to result in pGem-T Easy-ppm1Tb and subsequently subcloned as a 6.3 kbp EcoRI fragment into the HpaI site of plasmid pMV361-hyg (24) to result in pMV361-hyg-ppm1Tb. Complementation was confirmed by Southern blot analyses with a 0.2 kbp SmaI / NcoI lntMs upstream probe and a 0.2 kbp KpnI / HindIII ppm1Tb upstream probe.

Construction of expression vector pMV261-Gm-FusLppX
Plasmid pMV261-Gm a derivative of pMV261 is a shuttle vector replicating in E. coli as well as in mycobacteria (25). M. tuberculosis LppX was amplified by PCR from genomic DNA and fused to the M. tuberculosis 19 kDa promoter. Two sequences encoding a hemagglutinin and a hexa-His epitope were fused to the 3' part of the gene to facilitate subsequent purification and detection on Western blot and the insert was cloned into the EcoRI site to result in pMV261-Gm-FusLppX.

Preparation of cell extracts and Western blot analysis
Bacteria from 2 L cultures were harvested, resuspended in PBS containing Complete EDTA free tablets (Roche) to inhibit protein degradation and subjected to two French press cycles (American Instrument Company) at 2 × 10 8 Pa. Extracts were treated with 2 % Sodium Nlauroylsarcosine (SLS) for 1 h at room temperature and subsequently incubated at 4 °C over night. Soluble and insoluble fractions were separated by centrifugation at 30'000 g for 1 h at 4 °C. Extracts corresponding to 1-5 µg of total protein were separated by SDS-PAGE (12%) and analyzed by Western blot. Antiserum against HA epitope (Roche) was diluted 1:300.

FPLC protein purification
The soluble fraction of cell extracts was diluted with buffer containing 20 mM NaH 2 PO 4 , 0.5 M NaCl to 1 % SLS and loaded on HisTrap™ HP column (GE Healthcare) equilibrated with buffer containing 20 mM NaH 2 PO 4 , 0.5 M NaCl, 0.2 % SLS and 20 mM imidazole. Proteins were eluted with 0.125-0.5 M imidazole.

Mycobacterial lipoproteins are modified at the N-terminus
We chose the well characterized M. tuberculosis lipoprotein LppX (28) (Figure 1a). Purified LppX-HA-His from M. smegmatis parental strain was subjected to protein sequence analysis. Edman degradation of the prolipoprotein revealed a sequence starting at the initial methionine of the signal peptide of LppX ( Figure  1b). In contrast, no sequence was obtained from the mature LppX indicating a modification of the N-terminal amino group.

Identification of putative N-acyltransferases in bacterial genomes
In E. coli, N-acylation of lipoproteins is conferred by Lnt (29). We performed a BLAST search analysis (http://www.ncbi.nlm.nih.gov/sutil s/genom_table.cgi) with E. coli Lnt as a query to investigate the distribution of Lnt homologues in the bacterial kingdom and to identify putative homologues in mycobacteria. Lnt homologues are widely distributed in Gram-negative bacteria (α,β,γ,δ,ε Proteobacteria, Spirochetes, Aquifex, Cytophaga, Thermotoga), but absent from all classes (Clostridia, Mollicutes, Bacilli) of low GC Gram-positive bacteria (Firmicutes), although some indications for N-acylation in low GC Gram positives have been reported (12-15). In contrast, Lnt homologues were identified in all classes of high GC Gram-positive bacteria (Actinobacteria, e.g. Streptomyces, Nocardia, Corynebacteria and Mycobacteria) (Figure 2a), but Lnt activity of those homologues could not be demonstrated (22). The cell envelope of the phylum Actinobacteria is more complex than the cell envelope of Firmicutes. In M. tuberculosis and M. smegmatis, Rv2051c (Ppm1) and MSMEG_3860 (Ppm2) have the highest similarity to E. coli Lnt. M. tuberculosis Rv2051c encodes a two-domain protein, of which the N-terminal part shows similarity to E. coli Lnt. The C-terminal part of the protein encodes a polyprenolmonophosphomannose (Ppm) synthase, an enzyme involved in lipomannan and lipoarabinomannan synthesis (30). MSMEG_3860 has been shown to stabilize M. smegmatis Ppm1 in the bacterial membrane and therefore has been annotated as Ppm2 (31). MSMEG_3860 will be referred to as LntMs here. Lnt homologues are also present in Mycobacterium avium and Mycobacterium leprae and are encoded by a separate open reading frame as in M. smegmatis. The genomic region surrounding Lnt homologues is conserved in mycobacteria (Figure 2b).  (22). However, we could not restore growth of the PAP8508 mutant under restrictive conditions (data not shown). Seven amino acids (W237, E267, K335, E343, C387, Y388, E389) are reported to be essential for E. coli Lnt function (22). Five of these seven residues are conserved in LntMs, while two are altered (LntEc W237 corresponds to LntMs E323, LntEc Y388 corresponds to LntMs W477). We exploited site directed mutagenesis to introduce these E. coli codons into the M. smegmatis sequence of pUC18-lntMs to result in pUC18-lntMs323W, pUC18-lntMs477Y and pUC18-lntMs323W/477Y. However transformation of none of these vectors complemented the conditional E. coli lnt mutant (data not shown).

Generation
and characterization of M. smegmatis lnt::aph mutant Since we were unable to complement an E. coli lnt mutant, we decided to investigate Lnt activity directly in mycobacteria by generating a M. smegmatis lnt deletion mutant. The deletion mutant was constructed by transformation of M. smegmatis SmR5 with the suicide plasmid ptrpA1-rpsL-lntMs::aph using the rpsL counterselection strategy (20). The mutant strain resulting from allelic replacement is here referred to as M. smegmatis lnt::aph. Deletion of lntMs was verified by Southern blot analysis using a 5' lntMs DNA probe ( Figure S1). The probe hybridized to a 1.4 kbp fragment of the parental strain and a 6.4 kbp fragment of the lnt::aph mutant. The difference in size results from the deletion of a BstEII restriction site and insertion of a kanamycin resistance cassette. We cloned two complementation vectors (pMV361-hyg-lntMs and pMV361-hyg-ppm1Tb) expressing M. smegmatis Lnt and M. tuberculosis Ppm1 under control of their native promoters. Transformation of these plasmids into M. smegmatis lnt::aph mutant resulted in strains M. smegmatis lnt::aph-lntMs and M. smegmatis lnt::aph-ppm1Tb. Western blot analysis of extracts from M. smegmatis lnt::aph expressing LppX-HA-His revealed a molecular mass of the detected protein, which can not be distinguished from that of LppX-HA-His expressed in M. smegmatis parental strain. However, N-terminal sequencing revealed that LppX-HA-His purified from M. smegmatis lnt::aph is accessible to Edman degradation (sequence CSSP) indicating that the N-terminal amino group is not blocked anymore.

LntMs and Ppm1Tb are apolipoprotein Nacyltransferases
Since fatty acids of membrane phospholipids are the substrates for N-acylation of lipoproteins in E. coli (32)(33)(34), its lipoproteins are modified with myristic, palmitic, palmitoleic, oleic or vaccinic acid (35). Phospholipids in mycobacteria mainly consist of palmitic, palmitoleic, oleic and tuberculostearic acid (10-methyloctadecanoic acid) (36). Therefore we hypothesized that Nacylation of lipoproteins in mycobacteria increases the molecular mass by approximately 0.3 kDa. To differentiate between lipoproteins with a free or acylated N-terminus, we cloned an additional expression vector, RecLppX. It differs from LppX-HA-His by a hemagglutinin epitope followed by a thrombin cleavage site inserted after amino acid Ala (+19) of the mature LppX ( Figure  1b). The thrombin cleavage site LVPRGS was inserted to produce a small N-terminal fragment of 33 residues (about 3.5 kDa) after thrombin cleavage. To ensure that the insertion of a HAepitope and a thrombin site does not abolish recognition of RecLppX as a lipoprotein, we analyzed total lysates of M. smegmatis parental strain, M. smegmatis ΔlspA and M. smegmatis ΔlspA-lspA by Western blot (Figure 3a).
Temperature-sensitive lspA mutants of E. coli and lspA knock-out mutants of Gram-positive bacteria accumulate prolipoproteins (37,38). Immunoblotting of total lysates of M. smegmatis parental strain, ΔlspA and ΔlspA-lspA with antiserum against the HA-epitope (Roche) revealed the presence of a 25 kDa band in parental and complemented strains. In contrast, a band with a slightly larger size (appr. 27 kDa, the increase corresponds to the mass of the signal sequence) was observed in the ΔlspA strain. This result shows that the insertion of a HA-epitope and a thrombin cleavage site did not impair the recognition of RecLppX as a lipoprotein. We then investigated thrombin digested RecLppX for Lntdependent modification by Western blot analyses (Figure 3b). In the lnt::aph mutant we observed a slightly smaller size of the N-terminal part of RecLppX suggesting, that there are fewer modifications on the protein compared to parental strain and both complemented strains lnt::aph-lntMs and lnt::aph-ppm1Tb. In all strains, we also found a double band of the N-terminal part of RecLppX, indicating partial modification of RecLppX by other enzymes than Lnt or LspA. We also observed a deviation from the calculated size of the N-terminal part of mature RecLppX. The molecular mass was calculated to be 3.5 kDa but in the parental strain we found a band corresponding to a size of about 6 kDa. This difference in size is probably due to an altered migration behaviour because of lipid modifications. It can be excluded that these bands at 6 kDa are prolipoprotein forms, still containing the signal peptide, because the N-terminal part of pro-RecLppX from ΔlspA mutant is running at about 8.5 kDa. These results show that RecLppX is modified by LspA as well as by LntMs. Ppm1Tb is sufficient to replace LntMs implicating that similar lipoprotein modifications take place in M. tuberculosis. Recombinant M. tuberculosis LppX (FusLppX) was heterologously expressed and purified. Tryptic fragments of FusLppX were analyzed by MALDI-TOF/TOF mass spectrometry to characterize modifications taking place on lipoproteins in M. smegmatis at the molecular level. Purified mature LppX from parental strain, lnt::aph mutant and lnt::aph-ppm1Tb was prepared for analysis according to Ujihara et al. (27). For identification of the modifications of the universally conserved cysteine, the structure of the N-terminal tryptic peptide was determined. Experimentally found m/z values are summarized and compared to calculated m/z values in Table 1. Trypsin cleavage sites of LppX are given in Figure  1b Figure 5). In order to verify the diacylglyceryl modification, we also analysed tryptic peptides of LppX from the ΔlspA mutant by MALDI-TOF/TOF mass spectrometry ( Figure 6). Experimentally found m/z values are summarized and compared to calculated m/z values in Table 3. Trypsin cleavage sites of pro-LppX are given in Figure 1b. We found three m/z signals corresponding to the tryptic peptide containing the +1 cysteine. The signal at m/z = 4887.30 corresponds to the peptide with a disulfide bridge between the two present cysteines (position -8 and +1). The signal at m/z = 5041.23 corresponds to the peptide with both cysteines being modified by β-mercaptoethanyl, a buffer component used for SDS-PAGE. The signal at m/z = 5558.12 corresponds to the peptide with one cysteine being modified with a βmercaptoethanyl but the other being modified with the diacylglyceryl carrying O-linked C16 and C19:0 fatty acids also found in the previously analysed strains (Figure 6). This result shows that LppX purified from the ΔlspA mutant is a mixture of pre-pro-LppX and pro-LppX. Taken together the results show that the universally conserved cysteine of M. tuberculosis LppX is modified with a thioether linked diacylglyceryl residue carrying an ester-bound C19:0 and an ester-bound C16 fatty acid. In addition, it is modified with an amide-linked third C16 fatty acid. The C19:0 fatty acid corresponds most likely to the mycobacterial specific tuberculostearic acid. It is also proved, that LntMs is an N-acyltransferase and M. tuberculosis Ppm1 is able to complement M. smegmatis lnt::aph mutant and therefore Ppm1 seems to be a bifunctional protein. Within this protein the Nterminal domain presumably exhibits Nacyltransferase activity (our data) and the Cterminal domain exhibits mannosyl transferase activity (30).

DISCUSSION
The lipoprotein biosynthesis pathway consisting of the three enzymes Lgt, LspA and Lnt has been intensively studied in E. coli and has been shown to be essential and necessary for transport of lipoproteins to the outer membrane of Gramnegative bacteria (11,39,40). In mycobacteria, little is known about synthesis and localization of lipoproteins, only a few lipoproteins are functionally characterized and annotation is mainly based on theoretical considerations instead of experimental evidence. However, consistent with the biosynthetic pathway in E. coli, putative lgt (Rv1614) and lsp (Rv1539) genes have been identified in the M. tuberculosis genome (41). In revious studies we could show that in mycobacteria the lipoprotein pathway is a major virulence factor (3,4). For fundamental knowledge and further investigations, we were interested in how lipoproteins are modified in mycobacteria. In the present study, we investigated the lipid moieties of a representative mycobacterial lipoprotein. We identified Lnt homologues in mycobacteria, corynebacteria and streptomyces species. In low GC Gram positive bacteria Lnt homologues are completely absent (Figure 2), but in 1985 the first indirect detection of N-acylation in the Grampositive Bacillus subtilis was published and in S. aureus triacylation of lipoprotein SitC was recently reported (14,15), while another lipoprotein (SAOUHSC_02699) was only found to be diacylated (13). The protein responsible for attaching the third fatty acid to lipoproteins in S. aureus has not been identified. It may be differentially expressed depending on culture conditions or may have a narrow substrate specificity. In M. tuberculosis the Lnt homologue found, is annotated as Rv2051c. This ORF was originally annotated as a two-domain enzyme with a putative N-terminal Lnt domain and a C-terminal polyprenol monophosphomannose synthase (Ppm1) domain and was characterized as the latter one (30). Although the putative Lnt domain is not needed for Ppm1 activity, on overexpression in M. smegmatis it appeared to enhance the mannosyltransferase activity. Interestingly, the two domains of M. tuberculosis Ppm1 are encoded by separate, adjacent open reading frames in the genomes of other mycobacteria (Figure 2b).
Previous attempts to complement a conditional E. coli lnt mutant with Lnt homologues from other bacterial species corresponding to the order Actinomycetales (Streptomyces, Corynebacterium) failed (22). Likewise we were unable to complement this E. coli strain with a mycobacterial Lnt homologue. Even after exchange of the two essential amino acids differing between M. smegmatis and E. coli, complementation of E. coli lnt mutant failed. LntMs as E. coli Lnt attaches a C16 fatty acid to the free amino group of the universally conserved cysteine. Therefore the failure of complementation is not due to absence of fatty acid substrates. Rather mycobacterial lipoproteins are modified with a diacylglyceryl carrying mycobacterial specific fatty acids. Failure of complementation of PAP8508 therefore is probably due to the fact that LntMs recognizes only lipoproteins modified with a diacylglyceryl residue carrying at least one ester bound mycobacterial specific fatty acid. This implies that LntMs does not recognize lipoproteins modified with diacylglyceryl residues carrying only small fatty acids like palmitic or palmitoleic acid. Specificity could be tested in an in vitro assay system. Alternatively, the expression level or enzymatic activity of mycobacterial Lnt homologues may not sustain growth of fast growing E. coli. We then investigated LntMs and M. tuberculosis Ppm1 activity in a mycobacterial background. As lntMs is not an essential gene in mycobacteria, we generated an isogenic M. smegmatis lnt::aph mutant. After thrombin cleavage, the recombinant lipoprotein (RecLppX) extracted from M. smegmatis lnt::aph mutant showed a faster running behaviour on SDS-PAGE than RecLppX extracted from the parental strain. The size was about 0.3 kDa smaller corresponding to fewer modifications of RecLppX in the lnt::aph mutant. We also recognized a double band of digested RecLppX in all strains used as well as a discrepancy between the calculated and the apparent molecular mass. The altered running behaviour is probably due to the modifications on the small N-terminal fragment and the observed double band indicates partial processing of RecLppX by enzymes other than Lgt, LspA or Lnt. Glycosylation of RecLppX is one possibility, but information about the structure, function, and biosynthetic pathways of prokaryotic In this study we directly show, that Gram-positive mycobacteria synthesize triacylated lipoproteins. This is the first time to our knowledge that responsible genes for Lnt activity are assigned in Gram-positive bacteria. LntMs and M. tuberculosis Ppm1 are functional homologues of E. coli Lnt as they catalyze the transfer of the third acyl moiety to the free α-amino group of the N-terminal amino acid of lipoproteins. Most likely mycobacterial Lnt homologues differ in substrate specificity from E. coli Lnt. N-acylation is a prerequisite for transport of E. coli lipoproteins to the outer membrane (46). Likewise, N-acylation of mycobacterial lipoproteins may be required for transport to the outer most lipid layer of mycobacteria which according to recent investigations resembles the outer membrane of Gram-negatives (47,48).     palmitic acid α-thioglyceryl ester; palmitamide. Note that the ester-linked palmitic acid and tuberculostearic acid may be coupled to either position SN1 or SN2. Only one conformation is depicted.   (Figure 1b). Mass differences to the corresponding unmodified peptide (upper row) due to modifications are given in brackets. Observed modifications are: diacylglyceryl with a C16 fatty acid and tuberculostearic acid (C19:0) (+ 592.54 Da), plus eventually N-acyl with C16 fatty acid (+ 238. 23