Article pubs.acs.org/jpr
Investigation of the Listeria monocytogenes Scott A Acid Tolerance Response and Associated Physiological and Phenotypic Features via Whole Proteome Analysis John P. Bowman,* Esta Hages, Rolf E. Nilsson, Chawalit Kocharunchitt, and Tom Ross Food Safety Centre, Tasmanian Institute of Agriculture, University of Tasmania, Hobart, Tasmania, Australia S Supporting Information *
ABSTRACT: The global proteomic responses of the foodborne pathogen Listeria monocytogenes strain Scott A, during active growth and transition to the stationary growth phase under progressively more acidic conditions, created by addition of lactic acid and HCl, were investigated using label-free liquid chromatography/tandem mass spectrometry. Approximately 56% of the Scott A proteome was quantitatively assessable, and the data provides insight into its acquired acid tolerance response (ATR) as well as the relation of the ATR to the growth phase transition. Alterations in protein abundance due to acid stress were focused in proteins belonging to the L. monocytogenes common genome, with few strain-dependent proteins involved. However, one of the two complete prophage genomes appeared to enter lysogeny. During progressive acidification, the growth rate and yield were reduced 55% and 98%, respectively, in comparison to nonacidified control cultures. The maintenance of the growth rate was determined to be connected to activation of cytoplasmic pH homeostatic mechanisms while cellular reproductive-related and cell component turnover proteins were markedly more abundant in acid stressed cultures. Cell biomass accumulation was impeded predominantly due to repression of phosphodonorlinked enzymes involved with sugar phosphotransfer, glycolysis, and cell wall polymer biosynthesis. Acidification caused a shift from heterofermentation to an oxidatively stressed state in which ATP appears to be generated mainly through the pyruvate dehydrogenase/pyruvate oxidase/phosphotransacetylase/acetate kinase and branched chain acid dehydrogenase pathways. Analysis of regulons indicated energy conservation occurs due to repression by the GTP/isoleucine sensor CodY and also the RelA mediated stringent response. Whole proteome analysis proved to be an effective way to highlight proteins involved with the acquisition of the ATR. KEYWORDS: label-free, proteomics, multidimensional, Listeria monocytogenes, acid stress, growth phase
■
lifestyle within an animal or human host.7 Molecular approaches that measure gene expression, such as microarray analysis, provide a global view of cellular responses to stress. mRNAbased information alone, however, does not provide totally unambiguous information about biological processes occurring in response to stress within the cell because the relationship between gene expression (mRNA levels) and gene products (protein levels) only poorly correlates in microorganisms.8 The poor response coupling between mRNA and cognate proteins is due to several factors, including biologically manifested extrinsic noise associated with mRNA having low copy numbers,8 codon usage bias predicating mRNA expression intensity,9 posttranscriptional regulation and posttranslational modification,10 and differential turnover rates.11 Given this confounding complexity, proteomic profiles can potentially provide useful information about cellular functionality and manifested phenotypes. Knowing to what extent proteins are formed
INTRODUCTION Listeria monocytogenes is a robust Gram-positive pathogenic bacterium known for its ability to survive harsh environments, including the mild fermentation-based preservation processes used for many low pH foods, such as meat, cheese, orange juice, salad dressing, and yogurt.1 When exposed to adverse environments, such as acidic food, gastric secretions, or following phagocytosis, L. monocytogenes is particularly adept at initiating an acid-tolerance response (ATR) that allows it to survive more severe acidic conditions and cross-protects it against other forms of stress.2 Thus, exposure to conditions encountered prior to or during food fermentation-related processes, especially if the fermentation is slow, may allow survival of stress hurdles imposed in later stages of food manufacture. This survival phenomenon represents a risk during ready-to-eat food production, particularly products based on unpasteurized milk.3 The development of ATR in L. monocytogenes involves multiple genes controlled by a network of regulons.4−6 These systems are integral to its ability to occupy many environmental niches as a saprophyte and consequently switch to a parasitic © 2012 American Chemical Society
Received: November 17, 2011 Published: February 29, 2012 2409
dx.doi.org/10.1021/pr201137c | J. Proteome Res. 2012, 11, 2409−2426
Journal of Proteome Research
Article
Figure 1. Diagram showing the means by which L. monocytogenes Scott A was grown and sampled for protein analysis. Strain Scott A was grown statically in TSA-Ye broth at 30 °C, which was then acidified using DL-lactate and HCl pH 4.5. The graph shows growth during that time, lactate concentration, and pH. Sampling time points are also indicated. Two parallel experiments were performed. Control cultures were grown under the same conditions, except they were not acidified. The control cultures were sampled during the late exponential growth phase.
proteomics was performed here as a step toward robustly determining the physiological adaptation strategies used by L. monocytogenes to acidic environments. The same general proteomics approach we use here has previously been applied to examine the cell wall proteome of L. monocytogenes23,24 and also assess protein responses that reflect genetic and functional differences between strains.25 Though the broth system utilized is simple and does not attempt to replicate an actual food fermentation process, it includes relatively rapid acidification (using lactic acid and HCl to attain pH 5000 U/mg; Promega), at a sample protein to trypsin ratio of 40:1, at 37 °C with gentle shaking, overnight. The digestion was halted by addition of 30 μL of 10% formic acid. The samples were centrifuged for 5 min at 14000g to remove any insoluble material, and an aliquot (0.1−0.2 mL) containing 4 μg μL−1 of protein was transferred to HPLC vials for 2D-LC/MS/MS analysis.
Spectral Count Statistical Analyses
The nonlactic acid amended control culture was compared with the lactic acid amended cultures. Total spectral count normalization was used to normalize the data.33 Spectral counts from these samples were thus used to determine the fold-change in abundance between proteins across the time course experiment by using the approach developed by Old et al.18 A pseudospectral count of 0.5 was applied to each replicate.34 Validation of foldchange significance was performed using the beta-binomial distribution model.35 Derived significance values were also assessed by false discovery rate analysis36 with an α level of 0.05 considered significant for each individual time point compared to the control.
2D-LC-MS/MS
The separation of peptides utilized a nanoflow triphasic system consisting of a C18 capillary trap (Peptide CapTrap, Michrom BioResources, Auburn, CA) followed by a strong cation exchange resin (SCX, IntegraFrit Column, 100 μm i.d., 2.5 cm, New Objective, Woburn, MA) stage and a final analytical C18 nanocolumn (PicoFrit Column, 15 μm i.d. pulled tip, 10 cm, New Objective). Peptide samples were loaded onto the C18 capillary trap using a solution containing 0.1% formate and 0.1% trifluoroacetate (Merck). During sample loading, the SCX and analytical columns were switched via a valve out of line of the C18 trap, with the trap being washed to waste to ensure salts and other nonpeptide materials were not introduced. After sample loading, the SCX and analytical columns were switched in-line with the capillary trap and the flow was reduced via a splitter to 250 nL min−1. A five-step gradient process was performed followed by a 1.5 h column purge to prevent sample cross contamination (Supporting Information Data File 1). MS analysis utilized a electrospray ionization ion trap LTQOrbitrap mass spectrometer (ThermoElectron, San Jose, CA). During the MS/MS analysis, the peptides eluted from the microcapillary column were electrosprayed directly into the LTQ ion trap mass spectrometer with the application of a distal 2-kV spray voltage. The mass spectrometer was operated in the data-dependent mode to automatically switch between MS and MS/MS acquisition. A “survey” scan was performed in the electrostatic Orbitrap to identify the possible parent peptide masses, and the six most intense peaks were selected and fragmented in the ion trap, producing MS/MS data. This process was performed for 2 s and was cycled continually.
Clustering, Correlation, and Ontological Analyses
Proteins were classified within functional categories according to a scheme utilized within the ListiList database (genolist. pasteur.fr/ListiList/) adapted for transcriptomic studies22 as well as using DAVID (Database for Annotation, Visualization and Integrated Discovery) v. 6.7. Transporter proteins were classified into three separate subcategories: phosphotransferase systems (PTS), ABC-type transporters, and other transporter families. To determine changes in overall relative abundance in proteins within defined functional categories, normalized spectral counts were summed for all proteins detected. The TProfiler approach37 was used to assess differences between these relative abundance as adapted by Kocharunchitt et al.8 Statistical comparisons of functionally organized data sets were performed using nonmetric multidimensional scaling (nMDS) and Pearson correlation analysis using Primer 6 (Primer-E, Ivybridge, U.K.) visualize and compare the data on the basis of control and acidification treatments and also to determine trends in relative proteins abundance during the experiment. Temporal trends in relation to the controls were also investigated through use of unsupervised hierarchical clustering analysis utilizing Cluster 3.0.38
Peptide Identification and Bioinformatics
Peptide spectra were identified using the Computational Proteomics Analysis System (CPAS, LabKey Software Foundation) 2411
dx.doi.org/10.1021/pr201137c | J. Proteome Res. 2012, 11, 2409−2426
Journal of Proteome Research
Article
Multilocus Sequence Typing
MLST of Scott A was performed as described by Ragon et al.39 with sequence alleles matched against the Pasteur Institute Listeria MLST database (www.pasteur.fr/mlst). Fermentation End-Product Analysis
Scott A was cultured as indicated above and centrifuged. Liquid samples were analyzed by HPLC for aqueous metabolic endproducts in comparison with noninoculated TSA-YE broth using a HPLC Micromass ZP (Waters Corp., Milford, MA) fitted with a Waters Atlantis dC18 (19 mm × 150 mm, 10 μm) column and utilizing diode array UV−vis and evaporative light scattering detectors and a standard electrospray ionization mode. H2SO4 (5 mM) was used as the eluent at a flow rate of 0.3 mL min−1. Authentic chemical standards (Sigma-Aldrich) as well as mass spectral analysis were used to confirm end-product identity. Lactate levels were only effectively measured for the control cultures, since manually added lactate overwhelmed the levels of endogenously formed lactate.
■
Figure 2. Distribution of L. monocytogenes Scott A proteins identified in the study organized on the basis of cellular compartment (Supporting Information Data File 2) and by filtration criteria. To pass, the filter proteins had to have >95% confidence identification matches (Protein Prophet) and at least two unique peptides. Coverage consists of the average percent length of proteins covered by the peptide identifications. For proteins not detected and for proteins that passed or did not pass the filtration criteria the coverage is the proportion of proteins in the Scott A proteome that were not observed or observed, respectively.
RESULTS AND DISCUSSION This study utilized L. monocytogenes strain Scott A (ATCC 49594), a serotype 4b strain that was the causative agent of a listeriosis outbreak in Massachusetts, USA, in 198340 and since then is widely used in food-related research. MLST analysis placed Scott A in clonal complex 2, a common genetic subtype of L. monocytogenes,39 found to have worldwide distribution.41 In the acidification system used, the pH was manually changed over 8 h from pH 7.0 to pH 4.5, during which Scott A had a growth rate of 0.4 h−1 and attained a mean maximum density of 3.8 × 107 CFU mL−1. Growth ceased after 8 h, apparently due to nongrowth permissive acidity being attained (Figure 1). Shotgun LC/MS-MS analysis of cells at various time points (5, 7, 8, and 15 h) was used to infer responses to progressive acidification (Figure 1).
Growth Phase-Dependent and -Independent Proteomic Changes during Acid Stress
Proteins assigned to functionally allied sets (as shown in Figure 3B) were analyzed by both nMDS, Pearson correlation, and hierarchical clustering analysis to visualize changes in protein abundance within the overall data set. The results indicated that though responses to acid stress relative to the neutral pH control are large (Figure 4A), the analysis still proved sensitive enough to show a progressive temporal trend (Figure 4B) and clearly separates growth (5 and 7 h samples) and stationary growth phase states (8 and 15 h samples) (Figure 4C). Tolerance to mineral acid as well as to the presence of organic acids such as lactate and acetate is clearly increased in L. monocytogenes strains that enter the stationary growth phase,22 and on the basis of data here, proteins could be delineated that show distinct trends in having abundance being promoted or reduced during acid stress as well as being influenced by growth phase transition (Figure 4D, Supporting Information Data File 3). Many proteins exhibited dynamic abundance changes that are clearly growth phase-dependent, including 42 proteins exhibiting substantially greater abundance when growth ceases (Figure 4D). 84 proteins in the exponential growth phase have repressed levels of abundance but subsequently become more abundant with entry into the stationary growth phase. The ATR is clearly complex and incorporates adaptations that apparently begin in the exponential growth phase and that become more pronounced when growth stops. Strong negative correlations (Table 1) to acid stress were observed with cellular elements that relate directly to organic carbon acquisition, metabolism, and subsequent protein synthesis, including PTS transporters, enzymes of the central glycolytic pathway and secondary carbohydrate-related catabolic pathways, motility/chemotaxis, protein translationassociated proteins, and cold shock proteins. This negative
L. monocytogenes Scott A Proteome Is Dominated by Proteins Common to L. monocytogenes during Acid Stress
After normalization, 15475 spectral counts were obtained for the control and each individual time point. A total of 15023 unique peptides (combining all the samples analyzed) were identified, yielding a total 2371 protein identifications. Overall, 80% of proteins were detected from the Scott A genome; of these 1664 (56.3%) proteins passed identification filtration criteria (Supporting Information Data File 2). Checks against a reverse decoy database of all Listeria proteins revealed a low false identification error level (
