![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,
Uwe Kärst,1
Birgit Gerstel,1
Martin Loessner,2
Jürgen Wehland,1 and
Lothar Jänsch1*
Department of Cell Biology, Helmholtz Centre for Infection Research (HZI), Inhoffenstraße 7, D-38124 Braunschweig, Germany,1 Institut für Lebensmittel-u. Ernährungswissenschaften, ETH-Zentrum, Schmelzbergstrasse 9, CH-8092 Zürich, Switzerland2
Received 5 July 2006/ Accepted 4 October 2006
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
lgt mutant provided systematic insight into the relative expression of lipoproteins. Twenty-six of the 68 predicted lipoproteins were specifically released into the extracellular proteome of the lgt strain, and this proved that deletion of lgt is an excellent approach for experimental verification of listerial lipoproteins. Consequently, we generated lgt prfA double mutants to detect lipoproteins belonging to the main virulence regulon that is controlled by PrfA. Overall, we identified three lipoproteins whose extracellular levels are regulated and one lipoprotein that is posttranslationally modified depending on PrfA. It is noteworthy that in contrast to previous studies of Escherichia coli, we unambiguously demonstrated that lipidation by Lgt is not a prerequisite for activity of the lipoprotein-specific signal peptidase II (Lsp) in Listeria. |
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Surface proteins are attached to the bacterial envelope by several distinct mechanisms (5, 25). Among 132 predicted surface proteins of L. monocytogenes (35, 48), lipoproteins are the largest group of proteins with a common surface retention motif. Bacterial lipoproteins are characterized by the presence of specific signal peptides that are usually shorter than classical signal peptides and have a characteristic consensus sequence, referred to as the lipobox (commonly Leu–3-Ser/Ala–2-Ala/Gly–1-Cys+1) (37, 51). Lipoproteins are anchored to the outer surface of the cell membrane by a diacylglyceryl moiety, which is covalently bound to the invariant cysteine of the lipobox. The biosynthetic pathway of bacterial lipoproteins has been studied in detail using Brauns' lipoprotein of Escherichia coli (32). The key step for the subcellular localization of lipoproteins, lipidation, is catalyzed by the prolipoprotein diacylglyceryl transferase (Lgt). Lgt mediates the transfer of the diacylglceryl moiety from phosphatidylglycerol to the sulfhydryl group of the conserved cysteine. The same position is recognized by lipoprotein-specific signal peptidase II (Lsp), which cleaves the signal peptide within the lipobox (43). Analysis of the lipoprotein pathway in E. coli demonstrated that modification by Lgt is a prerequisite for Lsp activity, which suggests that a modified cysteine is required to coordinate the processes of maturation (13, 45). In gram-negative bacteria the N-terminal cysteine is further modified by addition of an amide-linked fatty acid by lipoprotein aminoacyl transferase (Lnt). As no orthologue of this enzyme has been found in the genomes of low-G+C-content gram-positive bacteria, this step in lipoprotein modification is obviously not conserved in all prokaryotes (22, 30).
In gram-positive bacteria lipoproteins perform various important roles in the milieu of the cell surface (for a review, see reference 40). They are key components of many transport systems, such as ABC transporters, but are also known to be involved in host-pathogen interactions (e.g., initiating inflammatory processes or facilitating adherence to eukaryotic cells) (1, 9, 18). The genome of L. monocytogenes encodes 68 putative lipoproteins comprising 28 substrate-binding components of ABC transport systems, 15 lipoproteins predicted to be involved in different enzymatic and metabolic activities, and, remarkably, 25 lipoproteins with unknown functions (12). Bioinformatic genome analyses have been shown to be a powerful tool for identification and functional assignment of putative lipoproteins (38, 39). However, pattern searches or web services created for prediction of bacterial lipoproteins either are based on relatively small data sets for experimentally verified lipoproteins (37) or focus on analyses of lipoproteins from gram-negative bacteria (Lipo at http://www.bioinfo.no/tools/lipo and LipoP at http://www.cbs.dtu.dk/services/LipoP/), although lipoboxes of gram-negative and gram-positive bacteria seem to be similar (42). Furthermore, the influence of the lipoprotein maturation process on the activity at the cell surface remains incompletely characterized and makes functional investigation of lipoproteins a challenging task, especially for gram-positive bacteria, in which lipoprotein expression is additionally masked under a thick peptidoglycan layer.
Reglier-Poupet et al. (30), who inactivated the gene encoding the lipoprotein-specific signal peptidase II of L. monocytogenes, demonstrated that maturation of lipoproteins is crucial for efficient phagosomal escape of the pathogen. However, this approach probably cannot provide information regarding the relevance of individual members of this protein family. Inactivation of proteolytic lipoprotein processing might result in surface accumulation of unprocessed proteins, which would affect bacterial growth rates and the capacity of bacteria to survive under hostile environmental conditions.
Here we report the effects of inactivation of the second enzyme involved in the lipoprotein pathway of L. monocytogenes, the prolipoprotein diacylglyceryl transferase (Lgt). We precisely defined the role of this enzyme in retention and translocation in comparison to the role of Lsp in Listeria, and in addition, we present evidence that casts new light on dogmas concerning bacterial lipoprotein processing. It is noteworthy that deletion of lgt did not affect the secretion of any nonlipoprotein in Listeria. Only lipoproteins were released along with normally secreted proteins into the extracellular compartment, and the expression of these lipoproteins could be analyzed systematically. To detect lipoproteins probably involved in the pathogenesis of L. monocytogenes and responsible for decreased intracellular replication of an lgt deletion strain, we used this strategy to detect the lipoproteins that are regulated by the virulence gene regulator PrfA. Proteome analyses of lgt strains that express PrfA differently revealed several regulated lipoproteins and indicated that there is PrfA-dependent posttranslational modification of the oligopeptide-binding protein (OppA).
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
prfA strain (8). The lgt strain, the complemented lgt(pE1lgt) strain, the lgt(pES11) strain, the lgt/prfA(pES11) strain, and the PrfA-overexpressing lgt(pEPS11prfa) strain were generated in the same genetic background (this study). Strains were grown in brain heart infusion (BHI) medium or minimal medium supplemented with 5 µg/ml erythromycin when they were carrying plasmids. Bacteria were cultivated in minimal medium as previously described (48). For inactivation of lipoprotein-specific signal peptidase II (Lsp), 20 mg globomycin (Sankyo Co., Tokyo, Japan) was dissolved in 1 ml of 70% (vol/vol) ethanol and added to BHI medium at a concentration of 100 µg/ml.
Construction of expression vectors pE1, pES11, and pEPS11. To obtain the pE1 expression vector, the pAT28 vector (47) was digested with XbaI and SphI. A new polylinker region (with restriction sites for NdeI, NcoI, NruI, ClaI, BglII, SalI, and XbaI) was inserted using oligonucleotides 5'-TCGAGCATATGCCATGGTCGCGAATCGATAGATCTGTCGACTGAGTAGGTAAT-3' and 5'-CTAGATTACCTACTCAGTCGACAGATCTATCGATTCGCGACCATGGCATATGCTCGAGCATG-3'. The linker was annealed and phosphorylated, which generated XbaI and SphI overlaps. It was cloned into the digested pAT28 vector. The resulting derivative (pLig46) was digested with NdeI and XhoI. A 0.23-kb fragment including the promoter region of actA was PCR amplified using L. monocytogenes total DNA as the template and primers 5'-TTTAATCCCATATGTACTCCCTCCTCG-3' and 5'-TGAAGCTCGAGAAGCAGTTGGGGT-3' containing sequences recognized by restriction endonucleases NdeI and XhoI (underlined). The amplification product was digested with NdeI and XhoI and ligated to pLig46. This generated a derivative (pLig57) which was used for integration of part of the lisA terminator region. For this pLig57 was digested with XbaI and cloned with a 85-bp fragment encompassing the terminator region of lisA, which was amplified using L. monocytogenes total DNA and primers 5'-AATAAAAAATCTAGAATAAAACCGCTTAAC-3' and 5'-GATAAACATCTAGATATTCTTTTACATTTG-3' containing sequences recognized by restriciton endonuclease XbaI at the 5' and 3' ends of the fragment (underlined). The fragment was digested with XbaI and cloned into pLig57. The resulting recombinant plasmid, pLig57b, was used as a cloning vector for the final expression vector, pE1. To obtain pE1, pLig57b was digested with XhoI and BamHI, and the resulting 0.5-kb fragment was cloned into the pCGU34 vector (27) digested with SalI and BamHI. For generation of pES11, the following primers were used for PCR amplification with L. monocytogenes total DNA as the template to amplify the promoter region and the signal sequence of actA: 3'-CTAGAATCTTCCATATGTGTCGCTGCAAA-3' and 5'-TGAAGCTCGAGAAGCAGTTGGGGT-3' (restriction sites are underlined). The 0.35-kb fragment was digested with NdeI and XhoI using a strategy analogous to the cloning strategy described above (using pLig57b as the cloning vector), generating a vector (pLig76) with the preferred Listeria codon usage of actA, as follows: 5'-CAT ATG CCA TGG TCG CGA ATC GAT AGA TCT GTC GAC TGA GTA GGT AAT CTA GAA-3'. The XhoI/BamHI fragment (approximately 0.5 kb) of this vector was transferred into pCGU34 digested with BamHI and SalI to generate pES11. To obtain pEPS11, pES11 was cut with KpnI and SacI. A 1-kb fragment encompassing the prfA gene was PCR amplified using L. monocytogenes total DNA as the template and primers 5'-AGCAACCTCGGTACCATAT-3' and 5'-CTGTTGGAGCTCTTCTTGGTGAAGCAATCG-3' containing sequences recognized by restriction endonucleases KpnI and SacI (underlined). The amplification product was digested and cloned into the KpnI and SacI sites of pES11, generating pEPS11.
Generation and complementation of the lgt deletion mutant.
The flanking regions of the lgt gene were amplified by PCR from the chromosome of L. monocytogenes wild-type strain EGDe. Primers Lgt1-F (CGTTGATTGGTTCTTCTCCGGCG) and Lgt2-R(Xba1) (GAACACCATTATCTAGAATTCCCCTAC) were designed to amplify a 459-bp fragment in the 5' flanking region of the lgt gene. Primers Lgt3-F(Xba1) (GGTAAAGTAGTTCTAGAGAAATAAAAAAGTTGG) and Lgt4-R (ATCTCTTCATATAAAGCACGAATCGC) were used to amplify a 358-bp fragment in the 3' flanking region of the lgt gene. After restriction with XbaI, the two fragments were ligated and cloned into the pCR-blunt 11-Topo vector using a TA cloning kit (Invitrogen, Karlsruhe, Germany). The plasmid was isolated from the recombinants and digested with HindIII and EcoRV, and the inserted DNA was cloned into the HindIII/SmaI-restricted pAULA vector. Following duplication of pAUL-A-lgt in E. coli TG2, pAUL-A-lgt was electroporated into L. monocytogenes wild-type strain EGDe, and gene replacement was performed as previously described by Schaferkordt and Chakraborty (34). The deletion was confirmed by PCR sequence analysis of chromosomal DNA from the lgt mutant. For construction of the prfA/lgt mutant pAUL-A-lgt was electroporated into an L. monocytogenes prfA mutant (8). Subsequent steps were carried out by using procedures analogous to the procedures used for construction of the lgt mutant. To complement the lgt mutation, lgt was amplified by PCR from the chromosome of the EGDe wild-type strain using primers Lgt5-F(Nde1) (GAGCATATGATGGGTAATGGTGTTCAGC) and Lgt6-R(Sal1) (GCGCGTCGACCTTCCTTTCTTAATCAAACTCG). The DNA fragment was digested with SalI and NdeI and inserted into SalI/NdeI-restricted shuttle vector pE1 downstream of the constitutive actA promoter. After duplication of pE1lgt in E. coli TG2, a PCR sequence analysis was performed to verify the sequence of the cloned gene. pE1lgt was transformed into the lgt strain by electroporation, and transformants were selected on BHI agar plates supplemented with 5 µg of erythromycin per ml.
Generation of polyclonal antibodies and immunoblot analysis. Polyclonal antibodies were prepared by immunization of rabbits with synthetic peptides. Peptides corresponding to residues 35 to 47 of the L. monocytogenes Lgt sequence, to residues 31 to 50 of the Lmo1800 sequence, and to residues 73 to 92 of the Lmo2595 sequence were synthesized with a terminal cysteine residue by the Department of Chemical Biology (GBF, Braunschweig, Germany). The peptides were conjugated with a maleimide-activated carrier protein using an Imject maleimide-activated immunogen conjugation kit (Pierce/Perbio, Bonn, Germany), and subsequent immunizations of rabbits were performed by Biogenes GmbH (Berlin, Germany). Immunoglobulin Gs were isolated by affinity chromatography on CNBr-activated Sepharose 4B (Amersham Biosciences, Freiburg, Germany). Lgt expression was analyzed by using protein extracts prepared from cultures of bacteria grown in BHI broth (optical density at 600 nm, 1.8; stationary phase). Proteins were separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) (21), and blotting was performed with polyvinylidene difluoride (PVDF) membranes in a semidry blot chamber as described previously (46) with a current of 1 mA/cm2 for 1 h. Lgt was detected with polyclonal antibodies (this study) (final dilution, 1:1,000) and peroxidase-coupled goat anti-rabbit secondary antibodies (Dianova, Hamburg, Germany). For analysis of Lmo1800 and Lmo2595 expression, different amounts of extracellular protein extracts were transferred to PVDF membranes using a slot blot unit. Protein detection was carried out with polyclonal antibodies (this study) (final dilution, 1:1,000) whose specificities had previously been tested by SDS-PAGE and immunoblot analyses of total protein extracts (data not shown). Chemoluminescent detection was performed with the Lumi-Light Western blotting substrate (Roche, Mannheim, Germany) and a cooled charge-coupled device camera (LAS-1000; Raytest, Straubenhardt, Germany). N-terminal sequencing was performed on a 494 Prouse protein sequencer (Applied Biosystems, Foster City, CA).
Metabolic labeling of lipoproteins with [14C]palmitic acid. Bacterial strains were cultivated in 20 ml of BHI medium supplemented with 10 µCi [14C]palmitic acid. Cells were harvested by centrifugation (10 min, 3,000 x g) in the late exponential phase and washed twice with phosphate-buffered saline (PBS). For generation of protoplasts, pelleted cells were resuspended in 2.5 ml of protoplast buffer (1 M sucrose, 30 mM NaCl, 10 mM MgCl2 · 6H2O, 50 mM Tris-HCl; pH 7.5) containing 0.33 mg/ml bacteriophage endolysin Hpl511 (23) and 0.01 mg/ml mutanolysin (Sigma-Aldrich, Taufkirchen, Germany). After incubation for 10 min at 37°C, 2.5 ml of MilliQ water and a protease inhibitor (Complete; Roche, Mannheim, Germany) were added. Protoplasts were disrupted by incubation on ice (10 min), followed by homogenization with a Potter glass homogenizer. After centrifugation at 10,000 x g (5 min, 4°C) to remove the cell debris and remaining bacteria, membranes in the supernatant were pelleted by centrifugation at 100,000 x g for 20 min (4°C). Membrane proteins were isolated by chloroform-methanol extraction as previously described (52). Proteins were separated by SDS-PAGE, transferred to a PVDF membrane, and exposed to a Fujifilm imaging plate (type BAS-MS 2025) for 2 to 3 weeks. 14C-labeled protein bands were visualized with a Fujifilm BAS-2500 phosphorimager (Raytest, Straubenhardt, Germany).
Preparation and two-dimensional (2-D) PAGE of extracellular protein fractions. Cultures (500 ml) of bacteria cultivated in ultrafiltered (molecular mass cutoff, 10 kDa) BHI or minimal medium were harvested by centrifugation (3,000 x g, 10 min, 4°C). The culture supernatants were filtered (pore size, 0.22 µm), and extracellular proteins were precipitated as previously described (48). The dried protein pellets were resolved in rehydration buffer {7 M urea, 2 M thiourea, 4% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), 30 mM dithiothreitol, 1.65 mM Tris, 0.5% IPG buffer (Amersham Biociences), protease inhibitor (Complete; Roche)}, and the protein concentration was determined as described by Bradford (4a). Isoeletric focusing was performed with IPGphor units using 18-cm immobilized pH gradients (pH 4 to 7) with a total voltage of 100 to 120 kV · h (0 V for 4 h, 30 V for 10 h, 30 to 300 V for 3 h, 300 to 3,000 V for 6 h, 300 to 5,500 V for 3 h, and 5,500 V until the end). The second-dimension analysis was carried out as described by Schaumburg et al. (35) using 12 to 15% gradient polyacrylamide gels. The gels were stained with RuBPS as described by Rabilloud et al. (29), recorded (LAS-1000 charge-coupled device camera; Raytest, Straubenhardt, Germany), and subsequently analyzed with ProteomWeaver 2.1 (Definiens, Munich, Germany).
Protein identification by mass spectrometry. For mass spectrometry, proteins cut from the 2-D gels were handled as described by Schaumburg et al. (35). Proteins were identified by peptide mass fingerprinting, as well as by using postsource decay fragmentation data recorded with a Bruker Ultraflex matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometer (Ultraflex; Bruker Daltonics, Bremen, Germany). The spectra were evaluated with the Biotools software (Bruker Daltonics) along with the MASCOT search engine (version 1.9; Matrix Science, London, United Kingdom). The criteria used to accept protein identifications included the extent of sequence coverage (minimum, 30%), the number of peptides matched (minimum number, 5), and the probability score (minimum Mowse score, 70). Proteins with lower scores were either verified manually or rejected.
In vitro infection analyses. Mouse fibroblast cell line 3T3 (= ACC 173; DSMZ) and human epithelial cell line Caco-2 (= ACC 169; DSMZ) were cultured in Dulbecco's modified Eagle's medium (DMEM) (1,000 mg/ml glucose; Gibco BRL/Invitrogen, Karlsruhe, Germany) supplemented with 2 mM L-glutamine and 10% fetal calf serum (FCS) at 37°C in the presence of 7% CO2. For the infection analysis, cells were grown to a density of 6 x 104 cells/cm2 in six-well tissue culture plates (Nunc) or (for microscopic studies) on glass coverslips (3 by 12 mm) in six-well plates. Monolayers were used for infection analyses 24 h (3T3) or 48 h (Caco-2) after seeding. Bacteria from 14-h cultures in BHI medium (stationary phase) were pelleted by centrifugation, washed once with PBS, and diluted in DMEM supplemented with 2 mM L-glutamine. Before inoculation of cells with bacteria at multiplicities of infection of approximately 6 bacteria per cell and 100 bacteria per cell (for microscopic studies), the cells were washed twice with PBS and overlaid with DMEM containing 2 mM L-glutamine but lacking FCS. Bacteria were centrifuged onto cells at 1,000 x g for 4 min. After 30 min of incubation, infected cells were washed twice with PBS and overlaid with fresh DMEM supplemented with 2 mM L-glutamine, 10% FCS, and 50 µg/ml gentamicin to kill extracellular bacteria. At selected times (1, 3, 5, and 7 h), ceIls were washed six times with PBS and lysed by addition of ice-cold 0.2% Triton X-100 for 1 min. Viable bacteria released from the cells were plated onto BHI agar plates to count colonies after 1 day. Each assay was carried out in triplicate and was repeated three times. Double fluorescence labeling of F-actin and Listeria was performed as described previously (8), using phalloidin coupled to Alexa Fluor 488 (Molecular Probes/Invitrogen, Karlsruhe, Germany) and polyclonal antibodies against L. monocytogenes revealed with Alexa Fluor 594-conjugated goat anti-rabbit (Molecular Probes). Samples were mounted and viewed with an epifluorescence microscope (Axiovert 135TV; Zeiss, Göttingen, Germany).
Bioinformatic lipoprotein analysis. A hidden Markov model (HMM) was constructed based on the 26 verified lipoproteins available. A multiple alignment was created with ClustalW using the shortest N-terminal sequence set (31 amino acids) that yielded an alignment in which all essential cysteines were lined up. The HMM was created using hmmbuild (HMMER package, release 2.3.2 [http://hmmmer.wustl.edu]). The complete genomic protein sequences of L. monocytogenes EGDe, Listeria innocua (12), and L. monocytogenes F2365 (26) downloaded from GenBank (accession no. AL592022.faa, AL591824.faa, and AE017262.faa) were then extracted with a BioPerl script to produce a FASTA-formatted file with the N-terminal 40 amino acids of only the proteins that had at least one cysteine between positions 15 and 30 of the sequence in order to remove sequences that could not be lipoproteins (the cysteine of the verified lipopoteins was at positions 18 to 23). These files were then analyzed by hmmsearch with the cutoff E-value set at 100.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
lgt) did not differ in cell and colony morphology and in in vitro growth analyses that were performed in BHI medium at 37°C (data not shown). The growth rates of the two strains were almost identical in the logarithmic phase. This observation indicated that Lgt is not essential for viability or cell division in rich medium.
Next, we characterized the lgt strain (i) under nutrient stress conditions and (ii) under more hostile conditions occurring during host cell invasion and intracellular growth. Cultivation in minimal medium resulted in slightly lower growth rates and a reduced final optical density for the deletion strain compared to the wild-type strain (see Fig. S3 in the supplemental material). To analyze whether lipoproteins and, in particular, their lipidation contribute significantly to the pathogenicity of L. monocytogenes, we exposed the nonprofessional phagocytic mouse fibroblast cell line 3T3 to the wild-type strain or the lgt mutant using a multiplicity of infection of 100 bacteria/cell. Three hours postinfection double staining with an anti-Listeria antibody and ß-phalloidin to visualize F-actin showed that the lgt mutant was generally able to invade and replicate within the mouse fibroblast cells. The mutant strain also formed actin tails similar to those formed by the wild-type strain, indicating that inactivation of lgt did not interfere with intracellular motility (Fig. 1A).
|
lgt mutant using a multiplicity of infection of 6 bacteria/cell. Cells were lysed at various times (1, 3, 5, and 7 h postinfection), and the numbers of intracellular bacteria were determined by counting colonies plated on BHI agar. No significant difference (P > 0.01, as determined by Student's test) between the wild-type and lgt mutant strains was observed 1 h postinfection with both cell lines (Fig. 1B). This observation indicated that lgt inactivation has no effect on bacterial entry into these cell lines. In contrast, 3, 5, and 7 h postinfection the number of intracellular lgt bacteria was significantly lower than the number of wild-type bacteria. For both cell lines the maximal relative reductions occurred 3 h postinfection, and the reductions were about threefold and twofold for the 3T3 and Caco-2 cells, respectively. At later times we observed smaller but still significant relative differences between the wild-type strain and the deletion mutant. In conclusion, the in vitro infection results demonstrated that prolipoprotein anchoring by Lgt contributes to intracellular growth of L. monocytogenes.
Lgt is exclusively responsible for lipidation of prolipoproteins in L. monocytogenes.
To ascertain that the effects described above were mediated exclusively by Lgt acting on lipoproteins, we constructed a complementation strain by transforming the lgt mutant with a multicopy plasmid harboring the wild-type lgt gene downstream of a constitutive promoter. The resulting strain was designated lgt(pE1lgt). To analyze expression of Lgt in the wild-type and complemented strains, we generated polyclonal antibodies. Using immunoblot analysis with the affinity-purified anti-Lgt antibodies, we detected increased expression of Lgt in the lgt(pE1lgt) strain compared to the wild-type strain. As expected, Lgt expression did not occur in the lgt mutant strain (Fig. 2A).
|
lgt mutant, the wild-type strain, and the lgt(pE1lgt) strain were cultivated in BHI medium in the presence of [14C]palmitic acid and harvested in the late exponential phase. To ascertain the specificity of the labeling, we first generated and purified protoplasts and then isolated membranes for a comparative analysis. Autoradiography of lipoprotein extracts separated by SDS-PAGE revealed several 14C-labeled protein bands in the samples of the wild-type strain and the complemented mutant (Fig. 2B). Although the level of Lgt expression was found to be increased in the complemented strain, the patterns and relative abundances of labeled proteins were remarkably similar to those of the wild-type strain. In contrast, no isotopically labeled protein was detected in the lgt strain, demonstrating that Lgt alone is responsible for the lipid modification of lipoproteins.
Deletion of lgt facilitates systematic identification and characterization of lipoproteins.
To investigate the effects of lgt inactivation on lipoprotein anchoring in L. monocytogenes, we examined the extracellular proteome of the lgt mutant by 2-D gel electrophoresis. Bacteria were cultivated in BHI medium, and proteins in the supernatant were obtained as previously described by Trost et al. (48). We used a pH gradient from pH 4 to 7 for the first gel dimension, as only 5 of the 68 annotated lipoproteins (Lmo1903, Lmo2349, Lmo0821, Lmo1379, and Lmo2854) have calculated isoelectric points greater than 7 (see Table S1 in the supplemental material). The extracellular expression pattern of the lgt mutant was strikingly different than that of the wild-type strain (Fig. 3A). Several spots were detected exclusively or were significantly upregulated in the extracellular proteome of the lgt mutant. Analyses of these spots by mass spectrometry resulted in identification of 24 different proteins. All of these proteins belong to the putative lipoprotein group, including various substrate-binding proteins and proteins involved in specific enzymatic activities, as well as several proteins with unknown functions (Table 1). The apparent pI and molecular mass values of the majority of the lipoproteins detected in the extracellular proteome of the lgt mutant correlate extremely well with the theoretical values calculated without the signal peptides. This observation suggested that signal peptides were cleaved off specifically and lipoproteins were released without further posttranslational modifications after inactivation of lgt. One exception was the Lmo2219 lipoprotein (Fig. 3A, spot 9), which was detected in two different spots whose pIs and molecular masses differed slightly.
|
|
lgt mutant. Both of these proteins are annotated as oligopeptide-binding proteins associated with ABC-type transport systems and have a particular role in oligopeptide transport for maintaining the physiological state of bacteria in rich media. Two other annotated di/oligopeptide-binding proteins were identified (Lmo0153 and Lmo2569), but their levels of expression were low compared to those of Lmo0135 and OppA.
Whereas inactivation of Lgt from B. subtilis resulted in release of different protein species into the supernatant (2), no additional or significantly increased protein species other than the annotated lipoproteins were detected in the protein pattern of the L. monocytogenes lgt mutant when this mutant was compared to the wild-type strain. Although our observations suggest that inactivation of lgt is an excellent strategy for systematic investigation of listerial lipoproteins, we found one lipoprotein, QoxA, whose surface retention was not impaired by the lgt mutation. QoxA was detected at comparable levels in the extracellular proteomes of the wild-type and lgt mutant strains, however, only in a spot whose apparent molecular mass was about 10 kDa lower than the theoretical value (Fig. 3A).
Comparative analysis of the extracellular proteomes of the wild-type and complemented strains demonstrated that expression of Lgt in the lgt mutant restored lipoprotein anchoring. It is noteworthy that no significant differences in the extracellular expression patterns were observed for either the lipoproteins or the other extracellular proteins (Fig. 3B). Even lipoproteins such as OppA and Lmo0135, which could always be found as part of the culture supernatant of the wild-type strain, were detected at comparable levels in the growth medium of the complemented strain. This analysis demonstrated that overexpressing Lgt has no detectable effect on the release of lipoproteins and other secreted proteins, thus underlining the specificity of its enzymatic function.
Retention, but not translocation, of listerial lipoproteins depends on Lgt.
To investigate the process of lipoprotein release in the lgt mutant, we performed N-terminal sequencing of the eight most abundant validated lipoproteins and the putative QoxA fragment. We found that QoxA has two N-terminal transmenbrane domains. The extracellular QoxA spot was cleaved immediately behind this region. As shown previously for B. subtilis (2), shedding of QoxA into the supernatant occurs independent of lipidation by Lgt. Individual N-terminal sequences could not be determined for OppA and Lmo0135; however, all other lipoproteins unambiguously had N-terminal sequences that were in perfect accordance with the cleavage site predictions for Lsp (Table 2). Only Lmo2219 was processed in two slightly different forms. Whereas the acidic form (Fig. 3A, spot 9) represented the perfect mature lipoprotein, the basic form was a homogeneous fraction of Lmo2219 starting with alanine at position –1. Apart from this peculiarity, Lsp seemed to be exclusively responsible for the processing of lipoproteins. Furthermore, lipidation by Lgt, which ensured the retention at the outer surface of the membrane, was obviously not an essential prerequisite for the Lsp-dependent processing of listerial prolipoproteins. This observation was surprising since in previous studies of other species workers concluded that Lsp can act only on diacylglyceryl-modified prolipoproteins (7, 13, 44).
|
lgt strain in the presence of globomycin, which specifically inhibits lipoprotein-specific signal peptidase II activity (7, 16, 33). We found that the release of the majority of lipoproteins into the supernatant was much impaired, resulting in a proteome pattern for the lgt strain that was remarkably similar to that for the wild-type strain (Fig. 4). Only Lmo0135 and Lmo2219 (basic form, starting with alanine) were still released into the supernatant. A comparison of these proteins with all other lipoproteins that were retained in the presence of globomycin did not reveal a biochemical property that would explain the different retention behavior. Thus, the majority of prolipoproteins most likely cannot complete their translocation across the membrane if Lsp is inhibited. In conclusion, Lgt activity is responsible for membrane retention, but it is not a prerequisite for signal peptide processing and translocation of listerial lipoproteins.
|
lgt strains with differences in PrfA expression and compared the lipoprotein expression profiles of these strains by 2-D gel electrophoresis. A prfA lgt strain analogous to the lgt strain was obtained, and a constitutive PrfA-overexpressing lgt strain was constructed by transforming the lgt mutant with a multicopy plasmid harboring the prfA gene of L. monocytogenes strain NCTC 7973 downstream of a constitutive promoter [lgt(pEPS11prfA)]. Since the possibility of additional effects on the extracellular proteome caused by this multicopy plasmid alone could not be excluded, we also transformed the lgt strain and the prfA lgt strain with the empty vector. The resulting strains were designated prfA/lgt(pES11) and lgt(pES11).
Bacteria were cultivated in minimal medium, since the expression of PrfA is generally reduced in a rich medium such as BHI medium (24). Growth analyses in minimal medium at 37°C revealed a slight growth advantage for the prfA/lgt(pES11) strain over the lgt(pES11) strain that most likely corresponded to the difference in transcription of virulence factors under PrfA-induced conditions. Consequently, growth of the constitutively PrfA-expressing lgt(pEPS11prfA) strain was delayed even further compared to growth of the other strains (Fig. 5). Hence, we harvested the prfA/lgt(pES11) and lgt(pES11) strains at different times, but always in the late exponential growth phase. The PrfA-overexpressing lgt strain was harvested at the same time as the lgt(pES11) strain at an optical density at 600 nm of 0.4, and its supernatant had a total protein content very similar to the total protein contents of both of the other strains. Protein secretion that is a prerequisite for the activity of many virulence factors is obviously markedly increased by overexpression of PrfA.
|
lgt(pES11) strain or whose levels were significantly increased in the extracellular proteome of the lgt(pES11) strain compared to the levels in the extracellular proteome of the prfA/lgt(pES11) strain. The levels of these proteins were further increased in the extracellular proteome of the lgt(pEPS11prfA) strain, demonstrating that there was a direct positive correlation with the amount of PrfA (Fig. 6). All regulated spots were analyzed by MALDI-TOF mass spectrometry, and the data resulted in identification of six different proteins. Three of these proteins were identified as the well-known PrfA-regulated proteins internalin (InlA), internalin C (InlC), and the actin assembly-inducing protein ActA. The other proteins were identified as the predicted lipoprotein Lmo0366 that was detected exclusively in the PrfA-overexpressing strain, Lmo2219, and the oligopeptide-binding protein OppA(1). Lmo2219 exhibits 45% identity with the posttranslocation chaperone PrsA of B. subtilis.
|
lgt/prfA(pES11) mutant, but its expression was reduced by concurrent expression of PrfA, as demonstrated with the lgt(pES11) and lgt(pEPS11prfA) strains. Mass spectrometric analysis unambiguously identified this spot as the oligopeptide-binding protein OppA. Its apparent isoelectric point differed markedly from that of the other OppA spot, which was in agreement with the calculated pI, pI 5.26 (without the signal sequence); thus, we designated this protein OppA(2). Since its apparent molecular weight is identical to that of OppA(1), the observed pI shift strongly suggests that there is PrfA-regulated posttranslational modification of OppA. No PrfA-dependent change in the total OppA protein level [OppA(1) plus OppA(2)] was detected.
To complement this comprehensive search for PrfA-regulated lipoproteins, we extended our analysis to low-abundance lipoproteins that could not be detected on the 2-D gels but may be involved in the pathogenic lifestyle of L. monocytogenes. We generated polyclonal antibodies against predicted lipoproteins Lmo1800 and Lmo2595, which attracted our attention because Lmo1800 is a putative tyrosine phosphatase, an activity associated with the virulence of YopA of Yersinia (10). Lmo2595 is a protein with an unknown function which has no orthologue in the closely related apathogenic species L. innocua. Surface anchoring of both predicted lipoproteins was significantly impaired by inactivation of lgt (Fig. 7A), which increased the number of experimentally validated lipoproteins to 26. To analyze whether these proteins are regulated by PrfA, we compared the extracellular proteomes of the different PrfA-expressing lgt strains by performing immunoblot analyses. Whereas Lmo1800 showed no PrfA-dependent expression, the extracellular amount of Lmo2595 was significantly decreased in all PrfA-expressing lgt strains, suggesting that there is negative regulation of Lmo2595 by PrfA (Fig. 7B).
|
Searching the L. innocua genome produced analogous results, 61 hits, removing Lin0800, the orthologue of Lmo0810, and Lin0626 with the cysteine at position 13 from the previous prediction. In addition, Lin1764 (unique to L. innocua) was predicted to be a lipoprotein. Previously, the genome of L. monocytogenes F2365 was predicted to contain 70 lipoproteins (reference 26, see Table S4 in the supplemental material). However, only 21 proteins are annotated as lipoproteins (UniProt) and another 5 proteins have an InterPro entry indicating lipid modification. Our HMM analysis identified 56 proteins as putative lipoproteins (see Table S2 in the supplemental material). Three other proteins annotated as "lipoprotein, putative" in this genome were predicted with only low significance. One of these, LMOf2365_2112, might well be a true lipoprotein with a divergent lipobox. In the case of LMOf2365_0173 and LMOf2365_2187 we regard the original annotation as probably incorrect (see Fig. S2 in the supplemental material).
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Lipoprotein generation by Lgt seems to be essential for growth in the gram-negative bacteria E. coli and Salmonella enterica serovar Typhimurium (11, 32). In striking contrast, inactivation of the lgt gene does not affect the viability of B. subtilis, Streptococcus pneumoniae, and Staphylococcus aureus (22, 28, 36). It is conceivable that in contrast to unmodified lipoproteins of gram-negative bacteria, unmodified lipoproteins of gram-positive bacteria retain at least part of their biological activity or are involved in less important functions. This hypothesis is supported by the results of investigations of the lipoprotein PrsA in B. subtilis, an extracellular chaperone which mediates the stability of exported proteins and thus plays an important role in protein secretion (17). Deletion of the prsA gene is lethal, indicating that PrsA is essential for growth of B. subtilis (20, 22). An lgt knockout mutant of B. subtilis is, however, fully viable, but its protein secretion is significantly impaired (22). Whereas inactivation of lgt had no visible effect on in vitro growth in rich medium, it clearly influenced intracytosolic growth of L. monocytogenes. This corroborates the importance of the results of our in vitro infection analyses, which demonstrated that selecting the correct growth conditions is pivotal for recognizing the real significance of gene functions. The lgt mutant was still able to invade nonphagocytic cells but always showed significantly reduced multiplication in the different cell lines. The fact that the relative differences between the wild type and the deletion mutant were maximal 3 h postinfection and not at later times might indicate either that lipoproteins have a specific role in the course of infection or that there are compensatory capacities of Listeria that warrant further study.
Besides lipoprotein anchoring maturation of prolipoproteins by the lipoprotein-specific signal peptidase II (Lsp) has also been demonstrated to affect intracellular survival of L. monocytogenes (30). Inactivation of the lsp gene resulted in a severe growth defect of L. monocytogenes in bone marrow macrophages, which was accompanied by a reduced capacity to escape from phagosomal compartments. Inactivation of lsp in Mycobacterium tuberculosis also led to markedly reduced virulence of this human pathogen (31), and deletion of lgt in S. pneumoniae resulted in decreased growth of S. pneumoniae in the respiratory tract of infected mice (28). As demonstrated by these examples, interfering with the lipoprotein pathway is a powerful strategy for analyzing the role of lipoproteins in the infectious processes of different bacteria. However, it provides no information about whether the alteration of a single lipoprotein or the cumulative effects of changes in the biological activity of numerous lipoproteins are responsible for the attenuated phenotype. In order to answer this question, it is necessary to examine which proteins are really lipid modified posttranslationally and thus directly affected by the targeted mutation.
Experimental and predictive characterization of listerial lipoproteins.
An increased release of lipoproteins into the extracellular proteome of a lgt mutant has previously been reported for B. subtilis (2). However, the extracellular levels of some autolysins were also increased in this mutant, demonstrating that inactivation of lgt also has an indirect effect on the release of nonlipoproteins. In contrast, we found no difference in secretion of nonlipoproteins between the lgt mutant and the wild-type strain of L. monocytogenes, although we improved our comparative proteome analysis by the use of narrow pH gradients in the first gel dimension (see Fig. S1 in the supplemental material). These data demonstrated that our comparative extracellular proteome analysis is a simple and very reliable approach for experimental verification of protein lipidation. Overall, we identified and verified 26 of 68 putative lipoproteins (12) specifically released into the extracellular proteome of the lgt mutant, making L. monocytogenes one of the best-studied species with regard to lipoproteins. The only limitation of our approach seems to be identification of lipoproteins having an additional surface retention motif. Indeed, apart from the QoxA fragment we found none of the five lipoproteins (Lmo1379, Lmo2125, Lmo2854, Lmo2184, and QoxA) having one or more predicted transmembrane domains in the extracellular proteome of the lgt mutant. This finding may also have been the result of low levels of expression that made these genes undetectable. Encouraged by the specificity of the experimental validation of listerial lipoproteins, we asked whether deletion of lgt can also be exploited for systematic expression analyses of these proteins. Deletion of lgt should cause the release of lipoproteins into the supernatant without affecting their translocation across the membrane. In E. coli different prolipoproteins accumulate at the inner membrane if the lipoprotein-specific signal peptidase II (Lsp) is specifically inhibited with the cyclic peptide antibiotic globomycin (13, 14). Prolipoproteins of E. coli cells treated in this way were detected on the outside of the inner membrane (15) and already had lipid modifications (13). From these data we concluded that (i) following translocation the prolipoproteins are retained at the membrane by their signal peptides and (ii) Lgt acts before Lsp. Indeed, inhibition of Lsp in our lgt deletion strain of L. monocytogenes by globomycin revealed that signal peptide processing is required to complete the transport across the membrane. In contrast, N-terminal sequencing of lipoproteins released from untreated cultures demonstrated that Lsp (at least in Listeria) can also process nonlipidated prolipoproteins. This is an important observation since it was supposed early that in E. coli lipidation strictly coordinates the proteolytic processing of lipoproteins (45), probably in order to ensure their retention at the bacterial membrane. An alternative (Lsp-independent) processsing by another protease has to be considered for PrsA, as suggested for B. subtilis (41), but is very unlikely to occur systematically. In fact, no alternative processing was observed for pre-OppA and pre-PrtM in Lactococcus lactis lacking the Lsp activity (50). Furthermore, none of the N-terminally sequenced lipoproteins in this study had the consensus sequence for proteolytic release that was recently proposed based on data obtained for B. subtilis (42). In conclusion, our data indicate that inactivation of the lgt gene in L. monocytogenes results in specific and unbiased release of lipoproteins into the growth medium. It is very interesting to look at the different functions of the verified lipoproteins. As expected, one-half of the identified lipoproteins belong to the group containing substrate-binding proteins, emphasizing the importance of lipoprotein metabolism in the uptake of different nutrients. We found considerable amounts of all of the predicted di/oligopeptide-binding proteins (OppA, Lmo0135, Lmo0153, and Lmo2569) in the extracellular proteome of the lgt mutant. It might be assumed that these proteins are expressed at different growth phases, depending on specific functions, such as sensing environmental changes via specific or nonspecific peptides. However, it is also possible that the proteins are expressed in parallel to guarantee an optimal nutrient supply. Proteome analyses of different growth phases based on this novel strategy should certainly help in studies of the mechanism of adaptation to environmental changes.
Our results for L. monocytogenes were used to define a genus-specific lipobox prediction (see Table S2 in the supplemental material). The HMM did not find all annotated putative lipoproteins (68 lipoproteins in L. monocytogenes EGDe) as its input is dependent on the proteins expressed under particular growth conditions, leading to a model that misses more divergent lipobox sequences. However, when the HMM was applied to the set of 33 gram-positive verified lipoproteins used by Sutcliffe and Harrington (37), all proteins were recognized, indicating that the HMM can be used for predicting lipoproteins in the phylum Firmicutes. It is particularly interesting that the verified lipoprotein Lmo1068 has orthologues in both L. innocua and L. monocytogenes F2365, but the essential cysteine of the lipobox is missing in the latter proteins. Furthermore, Lmo0460 and Lmo2595 have orthologues in F2365 but not in L. innocua, and Lin1377, the orthologue of Lmo1340 and LMOf2365_1357, has a cysteine at position 10 and was not accepted as a lipoprotein. These four proteins are candidates for further functional studies, because all four are either present or lipoproteins in the pathogenic organism L. monocytogenes but not in the apathogenic organism L. innocua.
Identification of lipoproteins regulated by PrfA. Consequently, we used our strategy to search for additional proteins regulated by PrfA, the major regulator of Listeria virulence gene expression. Overall, we found five proteins whose extracellular amounts were clearly increased in the presence of PrfA. Of these, we confirmed the known PrfA-dependent expression of the virulence factors internalin A and actin nucleating factor ActA, as well as the small secreted internalin C, which naturally occur in supernatant fractions of L. monocytogenes (48).
In addition to these expected nonlipoprotein virulence factors, we identified two lipoproteins whose expression positively correlates with the presence of PrfA. Lmo2219 is an orthologue of the posttranslocation chaperone PrsA of B. subtilis. Positive PrfA regulation of Lmo2219 has been demonstrated previously in vitro (24) and at the transcriptional level in vivo (6, 19). The other lipoprotein was identified as Lmo0366, a protein with an unknown function in which the absence of a PrfA box in the promoter region indicates that there is indirect PrfA regulation. In contrast to the known virulence genes, none of which are present in the genome of the closely related nonpathogenic species L. innocua, both Lmo2219 and Lmo0366 have orthologues in L. innocua. This indicates that Lmo0366, like Lmo2219, has a surface function which is not exclusively associated with the pathogenic lifestyle of L. monocytogenes but becomes more important during the infection process. The extracellular level of a third lipoprotein, Lmo2595, was significantly decreased in the presence of PrfA. Negative regulation by PrfA was reported previously; e.g., Milohanic et al. (24) identified eight genes which were downregulated parallel to increasing PrfA expression, but none of them is L. monocytogenes specific. It is noteworthy that Lmo2595 is present in the three sequenced L. monocytogenes genomes (EGDe, F2365, and H7858) but not in L. innocua, Listeria ivanovii (unpublished data), Listeria welshimeri (12a), and Listeria seeligeri (unpublished data). Therefore, this is the first example of a protein regulated negatively by PrfA and present exclusively in the genomes of L. monocytogenes strains. This underscores the assumption that downregulation of Lmo2595 might be important for the infection cycle of the human pathogen. However, as expression of lmo2595 was inconspicuous (factor 1) in the transcriptome analysis of Milohanic et al. (24), a posttranslational PrfA-dependent regulation process has to be postulated for Lmo2595.
One of the most striking differences between the lgt strains that express PrfA differentially strongly suggested a PrfA-dependent posttranslational modification of another downregulated lipoprotein. We detected two forms of the oligopeptide-binding protein OppA [OppA(1) and OppA(2)] with the same apparent molecular mass but significantly different isoelectric points (pI). With increasing PrfA expression OppA(2) disappears in favor of OppA(1). The PrfA-dependent modification of OppA indicates a specific role of the oligopeptide transport system during the infection cycle of L. monocytogenes. Indeed, deletion of the oppA gene was shown to result in delayed growth of L. monocytogenes in macrophages in vitro, as well as in organs of mice during the early phase of infection (4). A posttranslational modification of OppA might be associated with a different substrate-binding capacity or might regulate the binding to the corresponding ABC transporter complex. As the apparent pI of OppA(1) agrees with the calculated pI (pI 5.26) (without a signal sequence), it can be assumed that OppA(2) has the posttranslational modification. Phosphorylation is very unlikely, since preliminary results showed that the phosphospecific dye Pro-Q Diamond did not stain the two OppA forms differentially (data not shown). Therefore, a variety of possible posttranslational modifications now have to be considered and experimentally checked. The downregulation of oppA observed in in vivo transcriptomic studies (6, 19) probably did not depend on PrfA since the total protein level of OppA [OppA(1) plus OppA(2)] remained stable in our lgt strains that expressed PrfA differentially. Lmo0366 and Lmo2595 were not detected in these transcriptome studies.
In conclusion, our lgt deletion strategy permitted a detailed study of the lipoprotein synthesis pathway in gram-positive bacteria, demonstrated a lipidation-independent activity of Lsp, and produced a comprehensive list of validated lipoproteins. The fact that we found various lipoproteins regulated by the major regulator of Listeria virulence gene expression indicates that several lipoproteins, not a single lipoprotein, contribute to the pathogenicity of L. monocytogenes.
|
ACKNOWLEDGMENTS |
|---|
This work was supported by grants 031U207D and 0313313B from the "Verbundprojekt: Neue Methoden zur Erfassung des Gesamtproteoms von Bakterien" of the Federal Ministry of Education and Research.
|
FOOTNOTES |
|---|
Published ahead of print on 13 October 2006.
Supplemental material for this article may be found at http://jb.asm.org/.
Present address: Institut für Medizinische Immunologie, Martin-Luther-Universität, Halle-Wittenberg, Magdeburger Str. 2, D-06097 Halle, Germany.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
), and 
) strains in minimal medium at 37°C. OD600nm, optical density at 600 nm.
