|Home | About | Journals | Submit | Contact Us | Français|
The cell envelope of Mycobacterium tuberculosis (Mtb) is complex and diverse; composed of proteins intermingled in a matrix of peptidoglycan, mycolic acids, lipids, and carbohydrates. Proteomic studies of the Mtb cell wall have been limited; nonetheless, the characterization of resident and secreted proteins associated with the cell wall are critical to understanding bacterial survival and immune modulation in the host. In this study, the cell wall proteome was defined in order to better understand its unique biosynthetic and secretion processes. Mtb cell wall was subjected to extraction with organic solvents to remove noncovalently bound lipids and lipoglycans and remaining proteins were solubilized with either SDS, Guanidine-HCl, or TX-114. These extracts were analyzed by two-dimensional gel electrophoresis and mass-spectrometry and resulted in the identification of 234 total proteins. The lipoproteome of Mtb, enriched in the TX-114 extract, was further resolved by multidimensional chromatography and mass spectrometry to identify an additional 294 proteins. A query of the 528 total protein identifications against Neural Network or Hidden Markov model algorithms predicted secretion signals in 87 proteins. Classification of these 528 proteins also demonstrated that 35% are involved in small molecule metabolism and 25% are involved in macromolecule synthesis and degradation building upon evidence that the Mtb cell wall is actively engaged in mycobacterial survival and remodeling.
Despite efforts to eradicate tuberculosis, it remains one of the most successful and deadly human diseases. The tubercle bacilli resides within dendritic cells and macrophages, and in an immunocompetent host, the infection is walled off within a granuloma where it can remain dormant for years. While many factors contribute to its success, it is the thick, waxy cell wall of the bacillus that prevents dehydration, affords protection against varying levels of acidity and the detrimental affects of free radicals. In many respects, it is the unique architecture of the cell wall itself that makes Mycobacterium tuberculosis (Mtb) infection relatively difficult to treat with antibiotics. Since it is also a reservoir for many proteins and nonproteinaceous antigens, which are secreted into the extracellular milieu to stimulate and/or suppress the host immune response,1,2 its definition can be exploited for vaccine development.
For decades, the macromolecular features of the mycobacterial cell wall, including the mycolic acid and arabinogalactan core, have been studied in detail.(3) Structurally, the cell wall of Mtb is composed of a distinct inner core of mycolic acid-arabinogalactan-peptidoglycan (mAGP). In addition to the covalently attached lipids and carbohydrates, it is well-known that the free lipids, lipoglycans, and phosphotidyl inositols that reside in the outer core of the cell wall play key roles in modulation of the host immune response.(4) Specifically, the molecules, lipoarabinomannan (LAM), lipomannan (LM), and phosphatidyl inositol mannoside (PIM) are known to aid the process of host immune evasion.(2) In addition, virulence lipids such as trehalose dimycolate/monomycolate (TDM/TMM), phthiocerol dimycocerosate (PDIM), and sulfolipids (SL) and the protein machinery, such as MmpLs, required for their export are intercalated in the cell wall.(5)
Numerous cell wall associated proteins, including many lipoproteins and lipoglycoproteins, have also been described.6−8 For example, the TLR2 agonists, lpqH (19 kDa), pstS1 (38 kDa), and lprG (Rv14llc), are all found in the cell wall,7−11 where they function to regulate the action of macrophages and dendritic cells.(12) PstS1 also plays a role in bacterial escape from the host macrophage through apoptosis.(13) Many other lipoproteins of unknown function are identified in the secreted proteome of Mtb culture filtrate.14−16
The study of the secreted proteome of Mtb was driven by the search for novel immunodominant antigens, drug targets, and biomarkers for disease.17−20 To build upon this work, a number of studies employed advances in proteomic technologies such as two-dimensional gel electrophoresis (2DGE) and liquid chromatography mass spectrometry (LC−MS/MS) to further mine additional subcellular compartments of Mtb—cytosol, membrane, and cell wall.21−26 Historically, the proteins within the cell wall have been difficult to resolve and identify by traditional 2DGE methods.19,27 Mawuenyega et al., employing two-dimensional liquid chromatography coupled with mass spectrometry (2DLC/MS) were the most successful at defining the cell wall proteome with the identification of 306 proteins.(24) In this study, we set out to comprehensively describe the Mtb cell wall proteome in an effort to exploit additional proteins that may play a role in host-pathogen interactions and define new potential drug targets via discovery of unique biosynthetic or metabolic processes. We used a combination of detergent extraction, 2DGE, multidimensional liquid chromatography, and mass spectrometry to achieve our goal.
Mtb strain H37Rv was cultured in 2 L of glycerol alanine salts (GAS) medium(28) in roller bottles for 14 d at 37 °C with gentle agitation. Cells were harvested,(24) washed with phosphate-buffered saline (PBS), pH 7.4, and inactivated by γ-irradiation. Cells were disrupted in a French press, free lipids were removed, and the cell wall was obtained as previously described.(29) Briefly, 1 g of lyophilized cell wall was subjected to two extractions of 2 h each followed by one 18 h extraction with chloroform/methanol (2:1, v/v) at a ratio of 30 mL/g of cell wall. Extractions were performed at 22 °C with agitation. Centrifugation at 27000g for 30 min was performed to collect cell wall material. The 2:1 extracted cell wall was dried under N2 and further extracted twice for 2 h and one 18 h extraction with chloroform/methanol/water (10:10:3, v/v/v) each at 22 °C. The fully delipidated cell wall was dried under N2 and resuspended in PBS, pH 7.4. Cell wall protein was quantified by bicinchoninic acid (BCA) assay (Thermo Pierce).
Cell wall protein (CWP) was solubilized by one of three methods using (i) 6 M guanidine HCl (GuHCl), (ii) 2% sodium dodecyl sulfate (SDS), or (iii) 4% TritionX-114 (TX-114). For method (i), 75 mg of CWP was incubated at 22 °C for 4 h with agitation. The sample was centrifuged at 27000g for 30 min. Solubilized proteins were exchanged into 0.01 M NH4HCO3, and the protein amount was determined by BCA assay. The average protein recovery following GuHCl extraction of CWP was 40%. For method (ii), SDS soluble proteins were generated by extraction of 100 mg of CWP with 2% SDS in PBS (w/v) as described previously.(29) Briefly, the sample was bath sonicated for 3 h at 90 °C. The sample was centrifuged at 27000g at 22 °C for 30 min, collected, and centrifuged again. The fully cleared supernatant was then subjected to paired-ion extraction for removal of SDS.(30) Proteins were concentrated by centrifugation as above and pellets washed with acetone at −20 °C for 4 h. The final pellet was resuspended in 0.01 M NH4HCO3 and protein amount was determined by BCA. The average protein recovery following SDS extraction of CWP was 60%. In method (iii), a stock solution of 32% TX-114 in PBS was added to 300 mg of CWP to a final concentration of 4% detergent. Primary extraction occurred at 4 °C for 16 h. The extract was allowed to biphase at 37 °C for 30 min and was fully separated by centrifugation at 27000g for 30 min. Each phase was back extracted two additional times for 2 h each. The TX-114 detergent phases were pooled, and proteins collected by cold acetone precipitation.(31) Final TX-114 proteins were resuspended as above and protein amount was determined by BCA assay. Protein recovery following TX-114 extraction of CWP was 5%.
Samples of 200 μg of each CWP preparation were solubilized for 9−14 h in rehydration buffer (7 M urea/2 M thiourea, 1% amidosulfobetaine-14 (ASB-14), 1% 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS), 1 mM dithiothreitol (DTT), 0.25% NP-40, 0.625% ZOOM carrier ampholytes 3−10, and 1.9% ZOOM carrier ampholytes 4−7 (Invitrogen). Solubilized CWP was applied on ZOOM precast immobilized pH 4−7 linear gradient strips (7.0 cm; Invitrogen) according to manufacturer’s instructions. Focusing was achieved using a stepwise voltage gradient of 200, 450, 700, and 1000 V, for 10 min each followed by focusing at 2000 V for 2 h. SDS-PAGE of the isoelectric focusing (IEF) strips was performed using ZOOM 4−12% Bis-Tris SDS-PAGE gradient gels (Invitrogen). The gels were stained by Coomassie blue R-250 (Bio-Rad). Images were captured using a Gel-Doc System (Bio-Rad, Hercules, CA), and spot detection was performed using Delta 2D software (Greifswald, Germany).
A total of 5.0 mg of TX-114 CWP was digested with modified trypsin (Roche Diagnostics) at a ratio of 1:20 (E/S), in 0.1 M Urea and 20 mM methylamine. The digest was desalted using Sep-Pak Light C18 cartridge (Waters, Inc.) and concentrated under vacuum. The resultant peptides were separated by strong cation-exchange (SCX) chromatography using a polysulfylethyl A column (460 μm × 200 mm, 300 Å; Poly LC, Inc.) connected to a Waters Alliance analytical HPLC with a 2487 UV detector. The digest was applied in buffer A (5 mM K2PO4, 20% acetonitrile (ACN), pH 3.0) using a gradient of 0−80% buffer B (A with 0.5 M KCl) over 75 min with a flow rate of 1 mL/min. The elution of peptides was monitored at 214 nm, and fractions (3 mL each) were pooled based on UV absorbance. Each fraction (6 total) was concentrated under vacuum and resuspended into reverse phase buffer A (0.1% TFA in H2O). Individual SCX pools were subjected to further separation by reversed-phase HPLC (RP-HPLC) using a monomeric C18 column (4.6 mm × 150 mm, Vydac). Peptides were eluted using a gradient of 0−50% reverse-phase buffer B (90% ACN in A) over 40 min at a flow rate of 1 mL/min. Peptides were manually collected based on UV absorbance at 214 nm. A total of 139 RP-HPLC peptide fractions were collected and concentrated under vacuum.
Coomassie blue stained spots were excised from 2-DE gels and subjected to in-gel digestion32,33 with modified trypsin (Roche Diagnostics). Digests and 2DLC peptide fractions were resolved by liquid chromatography−mass spectrometry (LC−MS) using either an LCQ (2DGE) or LTQ (2DLC) as described previously.34,50 Tandem mass spectra were extracted, charge state deconvoluted, and deisotoped by BioWorks version 3.2 (Thermo Finnigan, San Jose, CA). All MS/MS samples were analyzed using Sequest (Thermo Finnigan; version 27, rev. 12) Sequest was set up to search the TB genome database (version 2.0, GenBank accession no. AL123456, 3912 entries) assuming the digestion enzyme trypsin, a fragment ion mass tolerance of 1.0 Da, a parent ion tolerance of 2.5 Da, and 3 allowable missed cleavages. Oxidation of methionine and acrylamide adduct of cysteine (2DGE spots) were specified in Sequest as variable modifications.
Peptide identifications were accepted if they exceeded specific database search engine thresholds. Sequest identifications required at least deltaCn scores of greater than 0.3 and XCorr scores of greater than 1.5, 2.2, 2.5, and 2.5 for singly, doubly, triply, and quadruply charged peptides (Supplementary Table S1).
Scaffold (version Scaffold-01_05_21, Proteome Software, Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 90% probability as specified by the Peptide Prophet algorithm.(35) Protein identifications were accepted if they could be established at greater than 90% probability as assigned by the Protein Prophet algorithm(35) and contained at least two unique peptides. Protein identifications at the lower threshold of acceptance were manually inspected for spectra quality. The protein probability false discovery rate for the CWP identifications was 3.5%.(36) Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony (Supplementary Table S1).
FASTA files for each annotated Mtb protein were collected and organized into small groups of 50. Each group was then submitted to the SignalP server (http://www.cbs.dtu.dk/services/SignalP/)37−39 using the gram-positive option. The output was saved in HTML format and a Perl script written to combine the results from all HTML files into one file and then converted to tab delimited format. This generated a complete list of results that were sorted in a spreadsheet to create the subsets of Neural Network positive, Hidden Markov Model positive, and the union of the NN and HMM sets. These lists were input into a java application along with the list of proteins isolated from the cell wall fractions. This application simply compared the predicted list to those isolated and identified the proteins found in the cell wall that were also predicted in each subset as having a signal peptide. To predict lipoproteins, the Mtb peptide fasta file from above was broken down into 1000 sequences per file. These were run through the downloadable version of LipoP (http://www.cbs.dtu.dk/services/LipoP/). The output files were combined in a spreadsheet, and from the sorted data, a list was extracted with every signal peptidase II cleavage site generating a predicted lipoprotein list. This list was input into a Java application that would compare the cell wall proteins, identifying the predicted lipoproteins from each cell wall fraction. The positive results verified that both Signal P models had predicted a signal peptide.
Protein preparations from the three CWP fractions resolved by 2DGE or SDS-PAGE were transferred to PVDF membranes and incubated with mouse monoclonal antibodies (α-GlcB [Rv1837c], α-PstS1 [Rv0934; IT-23], α-LprG [Rv1411c], α-Ald [Rv2780], α-LpqH [Rv3763; IT-19]; all available through the NIH, NIAID Contract, Tuberculosis Vaccine Testing and Research Materials) and polyclonal antibody against fructose bisphosphate aldolase (α-Fba [Rv0363c], kindly provided by Dr. Mary Jackson, Colorado State University) in TBS and 0.25% Tween 80 (TBST). After incubation, membranes were rinsed with TBST and developed using Amersham ECL Advance Western Blotting Detection Kit (GE Healthcare Life Science, Piscataway, NJ) as per manufacturers’ instructions. Protein immunoblots were visualized with the Typhoon 9400 multiplex fluorescent imager (GE Healthcare Life Science, Piscataway, NJ).
To maximize the resolution and protein identification of cell wall proteins, the delipidated cell wall of Mtb was subjected to extraction based on solubility within three reagents—an anionic (SDS), a chaotrophic (GuHCl), and a nonionic (TX-114) detergent. 2DGE of all CWP fractions demonstrated the majority of proteins resolving within a range of MW 25−75 kDa (Supplementary Figure 1). On average, 210 spots were resolved for GuHCl and SDS CWP fractions and 170 from TX-114. From the 2DGE analysis, 290 proteins were identified with 122, 131, and 37 proteins identified in the GuHCl, SDS, and TX-114 fractions, respectively. For all fractions, over half were unique to each subset (Figure (Figure1A).1A). Analysis of the data demonstrated 80% of proteins had known functions (Figure (Figure2).2). In contrast, the TX-114 extracted fraction was less amenable to gel separation most likely due to its increased hydrophobic content. A multidimensional chromatography approach40,41 was employed to better resolve proteins in this fraction, in order to exploit any unique or uncharacterized protein families within the TX-114 subset. While 2DGE was insufficient in identifying a significant number of proteins in the TX-114 detergent extract (Figure (Figure1A),1A), SCX chromatography combined with reverse phase chromatography of a digest of this CWP preparation allowed the resolution of 364 proteins, including an additional 294 proteins not found in any of the 2DGE preparations (Figure (Figure11B).
Validation of a few selected proteins, by Western blot analysis of samples resolved by 2DGE (Supplementary Figure 1) and SDS-PAGE (Supplementary Figure 2), corroborated the identification of these proteins by mass spectrometry.
All identified proteins were grouped by functional category as defined by Institute Pasteur, which demonstrated Mtb protein families present in each extract. Proteins in categories 3 (cell wall and cell wall processes) and 7 (intermediary metabolism) were consistently overrepresented among the CWP preparations (Figure (Figure2).2). Combining the data sets of both gel and non-gel based protein identifications led to the detection of 528 proteins (Supplementary Table S2). These were further classified into functional groups as defined by the Sanger Institute (Figure (Figure3A−C).3A−C). One hundred and five of the 528 cell wall proteins identified were not reported in previous Mtb proteome publications,16,19,22,24,27 including a comprehensive proteomic database http://web.mpiib-berlin.mpg.de/cgi-bin/pdbs/2d-page/extern/index.cgi,(19) and http://web.mpiib-berlin.mpg.de/cgi-bin/pdbs/2d-page/extern/index.cgi(42) (Table (Table11).
A majority of proteins were classified in either Category I, Small Molecule Metabolism (35%), or Category II, Macromolecule synthesis and degradation (25%) (Figure (Figure3).3). Subclasses of category I showed 19% classified in Small molecule metabolism-other (I.X), which included the classes: Central intermediary metabolism and Amino acid biosynthesis (Figure (Figure3;3; upper right); and subclasses of category II demonstrated an even distribution of proteins between synthesis/degradation of macromolecules (II.A, II.B, 12.5%) and cell envelope proteins (II.C, 12.1%) (Figure (Figure3;3; lower right).
The secreted proteins of Mtb have traditionally been characterized as important antigens and immune-modulators. Prior to being exported, many of these proteins are resident within the cell wall where their function remains largely unknown. To find the putative secreted proteins identified within the cell wall proteome, all identified proteins were subject to interrogation against Neural Network (NN) and Hidden Markov model (HMM) algorithms (SignalP, http://www.cbs.dtu.dk/services/SignalP/) which revealed 18%, 19%, 27% putative secreted proteins in GuHCl, SDS, TX-114 CWP samples, respectively, and 13% in the 2DLC resolved TX-114 CWP fraction. Cumulatively, of the 528 proteins identified in this study, 87 proteins were predicted to contain secretion signals and included the identification of 23 proteins uniquely found in this study (Table (Table2).2). A majority (60%) of the CWP associated with secretion are indeed within the category of the cell wall and its processes (Figure (Figure4).4). Further, many of these CWP are also membrane associated, either by description22,24,25 or functional annotation. Of the 23 secreted CWP unique to this study, 11 have been described to be involved with small molecule and peptide binding. Additionally, 23 of the 87 secreted proteins are classified as hypothetical or unknown (Categories V and VI of Sanger Institute) illustrating that to a large extent, the functions of these exported proteins are poorly understood.
Next, we interrogated the CWP to identify those with putative lipoprotein motifs. Here, a total of 16 proteins were predicted by LipoP (LipoP, http://www.cbs.dtu.dk/services/LipoP/)(43) to contain a signal peptidase II cleavage motif, and an additional 8 proteins were identified when compared to the 99 putative lipoproteins (2.5% of the genome) that reportedly exist within the Mtb proteome.(7) A majority of the lipoproteins contain no additional functional classification; however, 4 proteins (Rv0928, Rv0934, Rv2864c, Rv3666c) are involved with substrate binding and transport within the periplasm. One of two superoxide dismutases in Mtb, SodC (Rv0432), was also identified within the lipoprotein-enriched TX-114 CWP fraction. This protein has recently been defined as a highly glycosylated putative lipoprotein and is a defined B-cell antigen.44,45 Other putative lipoproteins identified in this study contain O-glycosylation motifs as predicted by NetOglyc and were found within the Mtb glycoproteome (Table (Table22).(46) The function of these dual-modified lipo-glycoproteins, as well as the nature of their modification remain largely undefined.
In this study, we focused on the gel-based two-dimensional separation of the cell wall and complemented that separation by liquid chromatography of digested TX-114 CWP. Identified protein families included known antigens, Fbp A, B, and C (Rv 3804c, Rv1886c, and Rv0129c), CFP10 (Rv3874) and Esat-6 like proteins EsxJ and EsxL (Rv1038c, Rv1198) along with antigens lpqH (Rv3763), lprG (Rv1411c), and lprA (Rv1270c). Others included members of the MmpL and Mce protein family of lipid export machinery. In concordance with previous work,26,47 a number of ribosomal proteins are reported here, which must indicate a high level of protein synthesis occurring at the cytosol−cell wall interface, most likely to facilitate the entry of proteins into the cellular envelope.
For over a decade, the advancement of techniques in 2DGE, mass spectrometry, and increased access to bioinformatic tools greatly enhanced proteomic studies of Mtb. Largely descriptive, these studies were undertaken to identify novel virulence factors and drug targets.18,19,27 Subsequent studies set out to understand functional relationships between proteins and discover antigens responsible for adaptive T-cell responses.23,24,26,48,49 The quantitative techniques of mass spectrometric profiling, isotope coded affinity tag (iCAT) and isobaric tagging for relative and absolute quantification (iTRAQ), have afforded the accurate monitoring of up- and down-regulation of proteins on a global scale.23,50 Previous work mining subcellular fractions of Mtb, focused on the cytosol and culture filtrate fractions, resulted in thorough characterization and the creation of 2D gel databases.19,42,51−53 Efforts have been made more recently to resolve the insoluble cell wall fraction. As mentioned previously, Mawuenyega et al. was able to identify 300 proteins within the cell wall using 2DLC as a separation method. Their study demonstrated functional relationships among key protein families within the fatty-acid synthesis pathway. Using these studies as experimental platforms, we were able to focus on the Mtb cell wall and were successful in reporting over 100 additional proteins that had not been identified previously. The elucidation of the cell wall proteome can facilitate subsequent studies of a more targeted nature where specific proteins can be monitored in response to various environmental, nutritional, or drug pressures.
One of the most interesting yet poorly understood protein families residing within the cell wall of Mtb are the triacylated lipoproteins. As stated above, Sutcliffe et al. predicted 2.5% of the genome to encode for lipoproteins. Defining these proteins beyond their antigenic and immune modulatory potential remains quite elusive with only a few studies predicting functional roles for Mtb lipoproteins.13,54−56 In this study, a vast majority of proteins found were involved in small molecule and macromolecule metabolism. Specifically, many of the CWP were annotated as small molecule and nutrient binding. This builds upon the evidence that clearly defines two Mtb CW lipoproteins, LppX and LprG, as aiding and/or facilitating the transport of lipids(57) and small molecules,(58) respectively. The presence of these lipoproteins and perhaps other secreted, CW resident proteins provides a conduit between various nutrient transportation pathways that are required for cellular survival and growth while allowing the mycobacterial cell wall to maintain its rigid hydrophobic integrity. While bacterial cell walls have historically been thought of as structural scaffolding with little active interplay between the extracellular milieu and inner cytosolic environment, this perception of a static bacterial structure is fading and we are beginning to understand the true complexity and dynamic nature of the biological processes being facilitated within the cell wall. For instance, these proteins may contribute to bacterial cell wall remodeling events in response to environmental stressors such as nutrient depletion, antibiotic pressures, the host immune response, and tissue remodeling events throughout the course of infection. For Mtb, the highly complex framework of the mAGP−outer lipid architecture can now be definitively complemented with a highly diverse protein population. This new perspective will provide further insight into cell wall remodeling processes during mycobacterial infection as well as give a more comprehensive list of potential drug targets and diagnostic candidates.
The cell wall proteins of Mtb were extracted and resolved by 2DGE or 2DLC. Several classes of proteins were identified, including putative secreted proteins, lipoproteins, and known T and B cell antigens. Many proteins have unknown function; however, their presence within the cell wall may be of biological relevance.
This work was supported by the NIH, NIAID contract HHSN266200400091c (K.M.D., PI). The authors thank Drs. John Belisle and Darragh Heaslip for their consultation and advice throughout the construction of this work and Dr. Carolina Mehaffy for technical advice in the writing of this manuscript.
National Institutes of Health, United States
Two tables, S1 and S2, are provided for validation of MS/MS data quality and complete listing of all proteins identified in this study, respectively. Figure S1 provides representative images of the 2DGE protein profiles for each CWP fraction. Figure S2 depicts Western blot validation of some proteins identified by MS/MS. This material is available free of charge via the Internet at http://pubs.acs.org.