Article pubs.acs.org/ac
Characterization of Residual Medium Peptides from Yersinia pestis Cultures Brian H. Clowers,* David S. Wunschel, Helen W. Kreuzer, Heather E. Engelmann, Nancy Valentine, and Karen L. Wahl Pacific Northwest National Laboratory, Richland, Washington 99352, United States S Supporting Information *
ABSTRACT: Here we demonstrate that when Yersinia pesitis is grown in laboratory media, peptides from the medium remain associated with cellular biomass even after washing and inactivation of the bacteria by different methods. These peptides are characteristic of the type of growth medium and of the manufacturer of the medium, reflecting the specific composition of the medium. We analyzed biomass-associated peptides from cultures of two attenuated strains of Yersinia pestis [KIM D27 (pgm-) and KIM D1 (lcr-)] grown in several formulations of 4 different media (tryptic soy broth (TSB), brain−heart infusion (BHI), Luria−Bertani broth (LB), and glucose (G) medium) made from components purchased from different suppliers. Despite the range of growth medium sources and the associated manufacturing processes used in their production, a high degree of peptide similarity was observed for a given medium recipe; however, notable differences in the termination points of select peptides were observed in media formulated using products from some suppliers, presumably reflecting the process by which a manufacturer performed protein hydrolysis for use in culture media. These results may help explain the presence of peptides not explicitly associated with target organisms during proteomic analysis of microbes and other biological systems that require culturing. While the primary aim of this work is to outline the range and type of medium peptides associated with Yersinia pestis biomass and improve the quality of proteomic measurements, these peptides may also represent a potentially useful forensic signature that could provide information about microbial culturing conditions.
T
what might be assumed to be pure microbial biomass. At the same time, their presence could reveal information about the growth environment of the cells. However, it has not been demonstrated whether growth medium peptides survive the culturing step and remain associated with the biomass in detectable concentrations. Because no single nutrient medium will support the growth of all types of bacteria, the scientific literature contains literally thousands of different medium formulations that serve varying purposes and microorganisms.2 In general, however, growth media contain the following elements: (a) an amino nitrogen source such as a protein hydrolysate or an infusion, (b) a growth factor source such as blood, serum, or yeast extract, and (c) an energy source, usually a sugar or other carbohydrate. Bacteria can use amino acids as sources of energy and carbon, and many common growth media do not contain added carbohydrates or sugars. Salts, trace metals, buffering agents, and selective agents can also be included. A few types of organic medium components are typically combined in various proportions and combinations to make the
he traditional proteomics community, while not explicitly ignoring low abundance peptides, has tended to focus on the abundant peptides in a sample when interpreting mass spectral peptide data. This emphasis is largely a consequence of the premium placed on protein identifications that arise from multiple, high confidence peptide-to-spectrum matches. Confident identification of low abundance peptides is typically difficult and severely complicated by the chronic under sampling of analytes within a proteomic sample.1 However, as quantitative proteomic comparisons as well as more thorough characterization of the proteins present become more important, more emphasis on less abundant peptides will become necessary. In proteomic analysis of cultured microbial or other cellular samples, it is often assumed that all observed peptides in the sample must originate from the cultured cells. However, the growth media used to culture bacteria often contain peptides that could render this assumption inaccurate. Most microbiological media contain nutritional components to supply amino nitrogen and to act as energy substrates for the cultured bacteria.2 These components often contain both free amino acids and peptides as products of protein hydrolysis. If residual medium peptides remain associated with the cultured cells, they could appear as confounding peptides in proteomic analyses of © 2013 American Chemical Society
Received: November 26, 2012 Accepted: March 14, 2013 Published: April 3, 2013 3933
dx.doi.org/10.1021/ac3034272 | Anal. Chem. 2013, 85, 3933−3939
Analytical Chemistry
Article
were handled in biological safety cabinets using Biosafety Level 2 precautions.9 The two strains were never cultured in the same laboratory or in different laboratories at the same time. Cultures were inactivated before any manipulations were performed on an open laboratory bench. Medium Formulations. Because medium components are manufactured by several suppliers who may use different processes and raw materials to make the components, it was expected that nominally identical medium components from different manufacturers might in fact vary in their peptide composition. For this study, we focused on four primary medium types for the growth of the KIM D27 (pgm) strain of Y. pestis: tryptic soy broth (TSB), brain−heart infusion (BHI), Luria−Bertani broth (LB), and glucose (G) medium. For experiments utilizing the KIM D1 (lcr) Y. pestis strain, TSB, BHI, and LB recipes were used. Tryptic soy broth typically contains hydrolysate of soy, hydrolysate of casein, and glucose. Brain−heart infusion broth commonly contains infusion of the named organs, meat protein hydrolysate, and glucose. Luria− Bertani broth combines hydrolysate of casein with yeast extract. Glucose medium contains yeast extract, glucose, and numerous salts. While rich in nutrient content, it should be noted that G medium is a suboptimal growth environment for Y. pestis due to an abundance of salts and an indeterminate range of amino acids. Nonetheless, this medium was used to explore the range of commonly available growth environments when culturing the KIM D27 strain. Because we obtained only minimal yields of biomass on G medium, despite multiple attempts, we eliminated this medium when culturing the KIM D1 strain. The medium formulations and detailed recipes can be found in the Supporting Information (Tables S1−S3). For this study, individual medium components from different manufacturers were used to assess the variability due to medium source, and several batches of each type of medium were made using components from different sources. Supporting Information Table S2 shows the sources and nomenclature of the various medium components, and Supporting Information Table S3 presents which components were used to formulate each test medium. Microbial Growth Conditions. A streak plate from a frozen stock of the desired strain of Y. pestis was made on tryptic soy agar (TSA). Following two days of growth at 30 °C and a check for purity by Gram staining, a single colony was placed into 2 mL of TSB medium and grown overnight at 30 °C to serve as a starter culture. Before initiating the growth of the larger culture, 100 μL of the starter culture was pelleted at 14 000 rpm for 1 min in a 1.5 mL microcentrifuge tube, the supernatant was removed, and the pellets were resuspended in 150 μL of the medium to be inoculated. For each formulation, 60 mL of medium was inoculated with 150 μL washed starter culture in a 250 mL flask and grown at 30 °C while shaking at 200 rpm. Triplicate cultures of Y. pestis KIM D27 were grown in each growth medium formulation shown in Supporting Information Table S3. Triplicate cultures of Y. pestis KIM D1 were grown in all media except G medium, TSB-2, TSB-4, TSB-5, and TSB-6. The optical density of each culture was monitored at 600 nm, and cells were harvested in the range of 0.6 to 1.4. Cultures were intentionally harvested across a range of optical densities to capture a variety of conditions to assess whether medium peptides were retained. A 15 mL aliquot of each culture was split into 3 centrifuge tubes for each medium formulation. After centrifugation at 6000 rpm for 15 min, the supernatant was
backbones of a wide variety of growth media. One of the most common types of medium component is a protein hydrolysate, also called a peptone, first described during the very early stages of modern microbiology.3 Peptones typically contain both free amino acids and oligopeptides, up to a molecular weight of approximately 6000 Da.4 The term peptone can be accurately used to refer to any water-soluble protein hydrolysate, but it is often used to refer specifically to hydrolysates of meat. The protein sources commonly used to make hydrolysates for microbiological media are meat, casein (the major protein in milk), or soy. Peptones are made by forming a slurry of the protein source and water, digesting the protein with either enzyme (typically trypsin, chymotrypsin, pepsin, papain, or a combination of enzymes found in porcine pancreas) or mineral acid, then purifying, and drying the resulting solution. While the reaction rates and enzymes may be variable, a measure of the amino nitrogen yield generally dictates the product end points. This process was pioneered and described in an oftenreferenced book chapter from 1970.4 Importantly, differences in the source or quality of the peptone have been previously shown to be important for the cultivation of pathogens.5 The differential impact on different types of pathogens led the authors to describe them as “pivotal components” of the culture medium. Another common type of medium component is an extract or infusion, which is made by extracting a raw material with water without hydrolysis. The raw materials are typically yeast, organs such as brain or heart, or meat. Extracts and infusions typically contain lower levels of peptides but higher levels of vitamins, trace metals, and complex carbohydrates than do peptones. An extract or infusion is often combined with a peptone in a culture medium. This combination of materials comprises a rich pool of potential sources of peptides within a microbiological medium. In an effort to determine whether growth medium peptides can be detected in association with harvested microbial biomass, we designed and executed a series of experiments using Yersinia pestis as a model system. Two different attenuated strains, KIM D27 (pgm-) and KIM D1 (lcr-), were utilized. These strains were cultured in four different growth media constructed from unique combinations of their basic components. In every case, we could detect medium peptides associated with the microbial biomass after harvesting, even when the biomass had been washed, autoclaved, and/or sterilized by treatment with bleach. Because these medium peptides persist through sample preparation, their presence has implications for the acquisition and interpretation of proteomic data from microbial samples.
■
EXPERIMENTAL METHODS Bacterial Strains and Biosafety. We used two attenuated strains of Y. pestis, a pgm and an lcr mutant. These two attenuated strains have been deemed exempt from regulation as select agents according to the National Registry of Select Agents and Toxins.6 Y. pestis KIM D27 is an isogenic derivative of KIM 10+ (biovar Mediaevalis) that was passaged until the pigmentation phenotype was lost due to spontaneous deletion of the pgm locus and associated virulence genes.7 Y. pestis KIM D1 lacks the virulence plasmid pCD1 responsible for the low calcium response.8 We verified strain identity with real time PCR assays that confirmed the absence of pCD1 in KIM D1, the absence of the pgm locus in KIM D27, and the presence of a marker specific to Y. pestis among the Yersiniae. Live cultures 3934
dx.doi.org/10.1021/ac3034272 | Anal. Chem. 2013, 85, 3933−3939
Analytical Chemistry
Article
the emphasis on trace peptide detection, blanks were inserted after each run to minimize carryover. In an effort to minimize complications arising from proprietary file formats, data from each LC-MS experiment was converted into mzXML.11 These files were further processed to minimize complexity based upon the procedure outlined by Renard et al.12 This secondary filter procedure minimizes low intensity contributions while maintaining pertinent spectral features. Data reduction prior to searching against a protein database using nonspecific enzyme rules greatly minimizes processing time. On the basis of a direct search comparison, the quality of the peptide spectrum matches (PSM) was not impacted when comparing the secondary processing with the raw data files (see Supporting Information, Figures 1S−3S). Unless specified otherwise, data sets were screened against a database composed of the nonredundant proteomes for the organisms most likely to be found in microbial growth media (Uniprot, www.uniprot.org). This included the protein sequences from cow, pig, soy, wheat, rice, and yeast along with the organism of interest, Y. pestis. For more detail on the database construction, please see the information within the Supporting Information. Data sets were screened against the medium protein database using X!Tandem and nonspecific enzymatic digestion rules due to the variability in mechanisms of protein digestion commonly used by medium manufacturers.13 The mass measurement accuracy for each parent ion scan was limited to ±10 ppm for data sets acquired using the Orbitrap mass spectrometer and +/300 ppm for scans acquired with the LTQ system. Given the method of preparation and environmental source of the peptides examined, a significant effort was spent on assessing the quality of each PSM. With this goal in mind, the output of each X!Tandem search was processed directly into a custom formatted text file compatible with the input requirements of the mass spectrometry generating function (MSGF) provided by Kim et al.14 This step was accomplished using a custom Python script that is available upon request. Peptide spectrum matches with MSGF scores less than 1 × 10−10 were excluded from consideration unless noted otherwise. It is worthy to note that this value is in accordance with the cutoff suggested within the Supporting Information of the aforementioned reference to limit false discovery rates below 1%.14
decanted and the cell pellet was resuspended in 15 mL of sterile phosphate buffered saline (PBS). At this stage, the resuspended cells were inactivated either by autoclaving at 121 °C for 20 min or by the addition of 10% bleach. Samples were stored at room temperature until preparation for proteomic analysis. This stage of culture processing was denoted as Wash 1. Unless specified otherwise, analysis and results were collected for Wash 1 samples. General microbiological preparation methods often specify a different number of washing steps depending on the desired outcome for the selected culture conditions. To further examine the range and quality of protein information as a function of wash cycles, Wash 1 for select cultures was subjected to 2 additional wash cycles (i.e., cell pelleting, supernatant decantation, and resuspension in PBS) producing Wash 2 and Wash 3. Isolation of Residual Medium Peptides. Because the focus of these investigations was on residual medium components, the procedure used to generate the samples for analysis was designed to minimize cellular disruption. The medium peptide extraction procedure began by mixing 100 μL of the homogeneous, inactivated cell suspension (e.g., Wash 1) with an equal part of acetonitrile (ACN, HPLC-Grade, Thermo-Fisher Scientific). This mixture was then placed in Thermomixer (Thermo-Fisher Scientific, Thousand Oaks, CA) and held at 45 °C for 60 min at 300 rpm. In order to prepare this solution for peptide extraction and purification, the ACN was evaporated while the sample volume was reduced to ∼50 μL using a Speed-Vac (Thermo-Fisher Scientific, Asheville, NC) and then reconstituted to 200 μL using 0.1% TFA (trifluoroacetic acid, Sigma-Aldrich, St. Louis, MO). Peptides were collected on PepClean C-18 columns (Thermo-Fisher Scientific (Formerly Pierce), Rockford, IL) according to the manufacturers instructions. Peptides were eluted from the cartridge by applying 20 μL of 90% ACN twice to maximize recovery. The eluted peptide solution was then reduced to ∼5− 10 μL and reconstituted to ∼25 μL using 0.5% FA (formic acid, Sigma Aldrich, St Louis, MO) prior to analysis using reversedphase liquid chromatography. Data Acquisition and Processing. Two μL of each concentrated sample was injected into a solvent system using a linear reversed-phase gradient with the following composition (100% A [5% ACN, 0.1% formic acid], B [95% ACN, 0.1% formic acid]) flowing at 2 μL/min and spanning 90 min. A linear gradient from 0% B to 60% B over 70 min was used to elute the sample peptides. The liquid chromatograph (Agilent 1200) was interfaced to a custom electrospray emitter10 using a 40 cm long capillary column (150 μm ID) packed in-house with C18 media possessing an average diameter of 5 μm (Phenomenx, Torrence, CA). Parent mass spectra were obtained using an Orbitrap XL mass spectrometer (ThermoFisher Scientific) with a mass resolution setting of 30 000. Tandem mass spectra for the top five most intense ions were acquired within the LTQ using default instrument settings related to normalized collision energy and number of microscans. To demonstrate that the method could be used on different types of mass spectrometers, certain data sets were acquired with a separate LC and LTQ mass spectrometer (Thermo, Thousand Oaks, CA). When appropriate, the distinction between the data sets derived from the different instrument platforms will be noted. Across the range of biological replicates and medium recipes, the sample order was randomized and shielded from the instrument operator. Given
■
RESULTS AND DISCUSSION
Compared to traditional proteomic efforts, the primary emphasis of this work was directed toward the characterization of peptides derived from the growth medium and requires a detailed account of the number and type of peptides observed across multiple samples. Commonly, there are a number of different means by which protein sources in microbial growth media may be prepared in order to provide a suitable set of nutrients for the target organism. Because these protein digestion mechanisms (e.g., enzymatic digestion and acid hydrolysis) are generally not rigorously controlled, the nature and composition of the resulting peptides and amino acids remain largely unknown. In addition, the uptake, consumption, and processing of medium peptides by the organisms could further alter the composition and state of the observed peptides. It is for these reasons that interpretation and confidence of the observed peptides must be assigned with care when focusing on residual medium peptides from microbial preparations. 3935
dx.doi.org/10.1021/ac3034272 | Anal. Chem. 2013, 85, 3933−3939
Analytical Chemistry
Article
For the medium formulations outlined in Supporting Information Tables S1 and S2, five primary sources of protein were present: tryptones, peptones, phytones, beef extract, and yeast extracts. The protein sources for TSB were approximately 73% by weight (see Supporting Information Table S-1) and were composed of various digests of casein and soybean meal, where the casein component was a minimum of 5-fold higher than the soybean content. In addition to being the most highly represented protein source by weight, the peptides in casein hydrolysates presumably all map to the casein protein, whereas peptides derived from a digest of soybean meal could be derived from a variety of proteins. It is therefore perhaps not surprising that the residual medium peptides for samples grown using TSB recipes were most highly correlated with different forms of casein (alpha, beta, etc.) with no observed peptide identifications to the soybean proteome. For BHI recipes, the total protein content was ∼74% by weight with approximately equal portions derived from beef heart, calf brains, and proteose peptones (derived from animal tissue). The residual peptides observed associated with cells from these BHI cultures were primarily derived from various forms of collagen (Uniprot: CO1A1_BOVIN, CO1A2-BOVIN) and connectin (Uniprot: Q28086 and Q28087). By weight, LB recipes contained the highest percentage of protein sources at ∼96%. The ratio of casein to yeast extracts in LB was 2:1. Casein again was the dominant source of residual medium peptides with a few peptides being derived from the yeast protein, glyceraldehyde3-phosphate dehydrogenase (Uniprot: G3P1_YEAST). At ∼29% protein content by weight, G medium contained the smallest amount of protein for the media examined. Yeast extract was the primary protein source in G media; however, no reproducible residual peptides were found associated with cells grown in G medium. Given the small size of the samples utilized, these peptides may well be below the level of detection. During the nonspecific digestion of proteins (either chemically or enzymatically), it is not uncommon for peptides that contain overlapping series of amino acids to be produced. As with genome assembly using overlapping sequence ladders, a similar approach may be employed for medium peptides using contiguous amino acid ladders to aid in the confidence of PSM.15 Shown in Figure 1 is a summary of observed peptide sequences for bovine casein across biomass from KIM D27 cultures from all of the formulations outlined in Supporting Information Table S3. The figure shows the composite protein coverage from the replicate cultures produced in each medium. The abscissa in Figure 1 corresponds to the amino acid index of bovine casein arranged from the N- to C-terminus (Uniprot ID: P02666), and each of the horizontal bars within the plot represents the identified peptide sequences. The gradation in the horizontal bars shown in Figure 1a captures the number of times a particular peptide was observed within a single data set. For the data sets shown, the median peptide observation frequency was 20. Because the enzymes or methods used for protein lysis by the various medium suppliers produce nonspecific digestion products, multiple overlapping peptide sequences were expected and observed. This peptide sequence overlap is shown in Figure 1b and summarizes the peptides observed for the TSB-6 sample between amino acids 72 and 107 of bovine casein. While the overlap between differing peptides approaches 1 between amino acid residues 82 and 85, the majority of this region maintains an average overlap of 5 ± 2. For horizontal bars that approach the black end of the
Figure 1. (a) Linear representation of observed peptide sequences for bovine casein across the range of selected medium formulations using Y. pestis KIM D27 cultures analyzed using Orbitrap instrumentation. The density of each bar represents the degree to which a given amino acid was observed. (b) Multiple, overlapping peptides are often observed for select regions of the protein sequence. These overlapping peptides are directly attributed to the nonspecific nature of the protein cleavage mechanism use during media preparation. This plot illustrates the range of overlapping peptides that comprise the densely covered amino acid sequence (72−107) found in TSB-6 and outlined in red shown in (a).
greyscale, a larger number of independent amino acid overlap is present with the opposite being true for the lighter end of the greyscale. The identity and origin of each biological replicate was hidden during preparation and analysis. Despite this randomization, a high degree of similarity was observed within each group of biological replicates. Casein peptides were observed in all samples that were known to contain casein in the medium recipe; however, in one instance, casein peptides with multiple degrees of amino overlap were observed in a sample not expected to contain this medium component (BHI-2). Brain− heart infusion medium recipes generally contain extracts from beef or pork with additional peptones included. In some cases, the peptones are from a defined source (e.g., gelatin) while others, such as proteose peptones, are generated from enzymatic digestion of ill-defined types of beef tissues. As a result, it is difficult to predict the protein source of the abundant peptides a priori in these components. On the basis of the information provided by the manufacturer for the particular proteose peptone used in this recipe, casein was not anticipated as a protein constituent of that component. After multiple replicates using alternating blank runs and analysis of the pure component, we concluded that this component did in fact contain β-casein. We could not determine whether this addition was intentional on behalf of the original component manufacturer or based on the information available from the supplier. Among the media examined, casein was a dominant protein when present in a formulation. Generally, a considerable degree of coverage was observed for this protein in the peptides that remained associated with harvested cells; however, under the examined conditions, only a few peptides were observed that corresponded to amino acid indices below 50. While this may be attributed to a number of factors including ionization 3936
dx.doi.org/10.1021/ac3034272 | Anal. Chem. 2013, 85, 3933−3939
Analytical Chemistry
Article
efficiency, degree of metabolic utilization by the organism, and general physiochemical properties, it is not surprising that peptides originating from amino acids 1−15 were not observed. Previous studies report that this region is a signal peptide which is usually cleaved following translation, making its observation using traditional mass spectrometry approaches less likely.16 Figure 2 illustrates the protein coverage and peptide overlap for the alpha-1 chain of bovine collagen (Uniprot ID: CO1A1)
Figure 2. Observed peptides for bovine collagen (CO1A1) from Y. pestis KIM D1 cultivated in BHI-1. Data was obtained with LTQ instrumentation. Using a similar representation as detailed in Figure 1, this plot illustrates a high degree of similarity between the different medium peptide sequences observed for BHI-1 across replicate injections.
Figure 3. Impact of successive cell mass washes on observed medium peptides from Y. pestis KIM D27 cultures using a LC-LTQ configuration for analysis. (a) Zoomed view of the peptides derived from bovine casein for Wash 2 and Wash 3 of the same sample (TSB3). (b) Remaining peptides for the alpha-1 chain of bovine collagen corresponding to Wash 2 and Wash 3 from BHI-1. In both cases, significant decreases in the total peptide coverage and number of observations occurred as a function of the number of wash steps.
obtained from replicate preparations of BHI-1 analyzed using an LTQ instrument. On the basis of the protein annotation, the primary chain of this protein spans from 162 to 1215 amino acids.17 For the same reasons as described for the signal peptide associated with casein, we only observed collagen peptides derived from the primary protein chain. Generally, only small differences in protein coverage were observed between replicate cultures, but occasionally, select peptides were observed in only one of the three replicates. This observation does not necessarily discount these identifications or minimize their significance (especially given the stringent filtering criteria used); rather, these low observation counts simply suggest a relatively low abundance of these peptides which would be expected to increase with larger sample amounts. Recognizing that spectral counting does not represent a means of absolute quantitation, this measure does provide a metric of relative abundance within the injected samples.18 In addition to the actual peptide concentration, variations in observed spectral counts are most likely due to the chemical and ionization characteristics of the given peptide. Previous efforts have demonstrated that the likelihood of peptide observation using LC-MS techniques is a function of multiple factors both physical and method related.19,20 To determine the effect of sample preparation on the range of peptides observed, a subset of the microbial cultures was subjected to additional washing steps prior to analysis. Figure 3 illustrates the peptides observed from the same volume of starting material (100 μL) from the second and third washes of KIM D27 cultures grown in BHI-1 and TSB-3. A zoomed view of the two protein sequences is provided as compared to Figures 1 and 2. It should be noted that, while there were small differences observed between the different strains (KIM D27 vs KIM D1), the magnitude of this variation did not exceed the
technical variation of sample preparation and analysis. However, for the two medium types examined, significant decreases in the total peptide coverage and number of observations occurred as a function of the number of wash steps. Nevertheless, for the most abundant proteins, trace levels of medium peptides remained in the final cell preparation. When possible, the use of larger amounts of cellular material could aid in the development of a more detailed profile of residual medium components. Under ideal circumstances, the ability to examine live, intact microbial systems is desired; however, health and safety considerations may preclude this possibility when the organism of interest is a human or animal pathogen. While there are numerous different mechanisms of microbial inactivation, we intentionally limited the focus of our initial examination of residual medium peptides to autoclaving and the use of bleach. Figure 4 demonstrates the relative levels of amino acid coverage observed for bovine casein between the autoclaved and bleached systems (a) and (b), respectively. As the figure illustrates, a large degree of similarity exists between the conditions though notable differences related to abundance of the observed species are evident. Efforts were made to identify unknown chemical modifications that could have accounted for these abundance differences using a method similar to Menschaert et al.; however, no discernible patterns were observed (data not shown).21 This initial exploration into the differences as a function of inactivation condition does not preclude the possibility that chemical modifications occurred. Future work will need to address this possibility and others (e.g., peptide degradation) which would shift experimental data outside traditional database search parameters. 3937
dx.doi.org/10.1021/ac3034272 | Anal. Chem. 2013, 85, 3933−3939
Analytical Chemistry
Article
medium. Although outside the scope of this work, these observations beg the question of whether their absence is due to direct consumption or transformation by the microbial system or the result of chemical bias during extraction and purification. Nonetheless, the range of peptides detected in the sterile media directly informs, at a minimum, the range of species that may be detected in the final microbial product. Compared to traditional bottom-up proteomics that utilize highly standardized enzymatic digestion for peptide production, media suppliers and traditional microbiological protocols utilize protein-based nutrient sources that are often digested in a nonspecific fashion. Because culture reproducibility is desirable, it is not surprising that the range of proteins observed in microbial media is relatively limited. While this may not be true for all systems, this generalization appeared to hold across this investigation. This consistency may be due to a number of factors related to manufacturing or production methods; however, more interesting was the high degree of reproducibility observed for the peptides present. Because nonspecific digestion is commonly used when developing microbial growth media, it was expected that peptide fragments across the entire protein sequence would be observed. As many of the figures presented demonstrate, the exact opposite was observed; specific ranges of peptides were found across all media conditions. The exact physiochemical mechanism by which these peptide fragments are retained through the manufacturing and growth process is unknown; however, their consistent observation in microbial preparations should warrant their consideration when analyzing data from proteomic samples derived from cells grown in a medium that itself was derived from a biological source. Granted, most proteomic searching approaches confine results to a range of cleavage sites; however, a closer examination of the peptides found in sterile media (Supporting Information Table S4) revealed that nearly half of the observed peptides were semitryptic in nature, making their account especially relevant to the quantitative proteomics field.
Figure 4. Comparison of inactivation condition on the observed peptide ranges for bovine casein derived from Y. pestis KIM D1 cultures as determined using an Orbitrap system. (a) Represents the amino acid ranges observed for autoclaved samples while (b) illustrates similar profiles for bleach inactivated samples.
Another relevant question central to this effort is how well the peptide content found in the sterilized media correlated with observed peptides in the final preparations. A representative, sterilized aliquot of the different medium types was examined using the outlined protocol. This exercise demonstrated that all of the residual peptides observed associated with cell biomass after growth were also found within the sterilized media prior to introduction of the bacteria. Almost all of the peptides observed in the sterilized media were overlapping in nature (similar to Figure 1), reflecting the nature of the medium component manufacturing processes. The peptides present in sterile recipes of LB, TSB, BHI and G medium could be associated with a limited range of annotated proteins. A full account of these proteins and corresponding peptides can be found in Supporting Information Table S4. To briefly summarize, in the G recipes, only a few peptides mapping to yeast glyceraldehyde-3-phosphate dehydrogenase (GAPDH, P00360) were observed, which reflects the comparatively low protein concentration in this medium. For the sterile BHI recipe, peptides of collagen and connective tissue were most commonly observed; however, lower levels of manufacturer enzymes used for digestion of the original protein sources were also seen. In the media containing appreciable protein and peptide content (specifically, TSB, LB, and BHI), approximately 30 different peptides mapping to a range of unknown proteins from cow, pig, rice, and soy bean sources were also observed. Their consistent observation (Supporting Information Table S4) demonstrates their relevancy as proteins, but they have simply not been annotated within the protein databases that represent possible medium components. In the sterile TSB preparations, approximately 79% sequence coverage of bovine casein was observed compared to 34% coverage for the residual peptide subset found in the final microbial preparation. As this comparison demonstrates, a sizable fraction of peptides in the sterile medium recipe did not pass through to the final preparation. Most notably absent in the residual peptide preparations were peptides related to soybean (Glycine max), rice proteins (Oryza sativa), and some of the lower abundance connective proteins from animal sources (see Supporting Information Table S4). However, the lack of observation in the residual preparations is not entirely surprising given the disparate levels of proteins in the sterile
■
CONCLUSIONS By utilizing a highly conservative measure of PSM confidence, we have demonstrated a tailored analytical approach to characterize the residual medium peptides associated with Y. pestis cells cultured in a range of media. This sample handling and analytical procedure is geared toward maintaining cellular integrity while extracting the maximum amount of protein information related to the microbial growth conditions used during cultivation. Information about production conditions could potentially be useful in microbial forensic investigations, and the peptide identities associated with biomass could also constitute a forensic signature. Our initial investigations into other organisms including Burkholderia thailandensis and Bacillus anthracis grown in TSB and BHI, respectively, also show similar residual peptide profiles. While outside the scope of this work, future efforts aim to utilize the subtle differences between different peptide signatures to provide a finer degree of differentiation between microbial growth conditions. The presence of these medium peptides also has distinct implications for interpretation proteomic data from microbiological samples. Though many of the peptides observed in this work did not fall under the tryptic category, many of them were semitryptic in nature and would contribute to any initial assessment of total protein and peptide content of the microbial biomass using traditional assay methods. Armed with the knowledge that residual growth medium peptides are retained 3938
dx.doi.org/10.1021/ac3034272 | Anal. Chem. 2013, 85, 3933−3939
Analytical Chemistry
Article
(18) Neilson, K. A.; Ali, N. A.; Muralidharan, S.; Mirzaei, M.; Mariani, M.; Assadourian, G.; Lee, A.; van Sluyter, S. C.; Haynes, P. A. Proteomics 2011, 11, 535. (19) Craig, R.; Cortens, J. P.; Beavis, R. C. Rapid Commun. Mass Spectrom. 2005, 19, 1844. (20) Mallick, P.; Schirle, M.; Chen, S. S.; Flory, M. R.; Lee, H.; Martin, D.; Raught, B.; Schmitt, R.; Werner, T.; Kuster, B.; Aebersold, R. Nat. Biotechnol. 2007, 25, 125. (21) Menschaert, G.; Vandekerckhove, T. T. M.; Landuyt, B.; Hayakawa, E.; Schoofs, L.; Luyten, W.; Van Criekinge, W. Proteomics 2009, 9, 4381.
through a microbial preparation, researchers may take steps to minimize quantitative errors.
■
ASSOCIATED CONTENT
S Supporting Information *
Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors would like to recognize Aaron C. Robinson for his assistance in managing and curating the experimental databases used in this study. Funding for this work was provided in part through contract AGRHSHQDC07X00451 to Pacific Northwest National Laboratory by the Department of Homeland Security Science and Technology Directorate. Additional funding was provided by the laboratory directed research and development (LDRD) program at Pacific Northwest National Laboratory. Battelle Memorial Institute operates Pacific Northwest National Laboratory for the U.S. DOE under Contract DE-AC06-76RLO.
■
REFERENCES
(1) Michalski, A.; Cox, J.; Mann, M. J. Proteome Res. 2011, 10, 1785. (2) Atlas, R. M. Handbook of microbiological media; Taylor and Francis: Boca Raton, FL, 2010. (3) Naegeli, C. Sitz’ber, math-physik, Klasse Akad. Wiss.: Muenchen; 1880, 10, p 277. (4) Bridson, E. Y. Rev. Med. Microbiol. 1990, 1, 1. (5) Gray, V. L.; Muller, C. T.; Watkins, I. D.; Lloyd, D. J. Appl. Microbiol. 2008, 104, 554. (6) CDC. http://www.selectagents.gov/select agents and toxins exclusions.html. Accessed March 19, 2013. (7) Une, T.; Brubaker, R. R. Infect. Immun. 1984, 43, 895. (8) Meyer, K. F.; Cavanaugh, D. C.; Bartelloni, P. J.; Marshall, J. D., Jr. J. Infect. Dis. 1974, 129 (Suppl 1), S13−S18. (9) Chosewood, L. C., Wilson, D. E., Eds. Biosafety in microbiological and biomedical laboratories, 5th ed.; U.S. Dept. of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Institutes of Health: Washington, D.C., 2009. (10) Kelly, R. T.; Page, J. S.; Luo, Q. Z.; Moore, R. J.; Orton, D. J.; Tang, K. Q.; Smith, R. D. Anal. Chem. 2006, 78, 7796. (11) Pedrioli, P. G. A.; Eng, J. K.; Hubley, R.; Vogelzang, M.; Deutsch, E. W.; Raught, B.; Pratt, B.; Nilsson, E.; Angeletti, R. H.; Apweiler, R.; Cheung, K.; Costello, C. E.; Hermjakob, H.; Huang, S.; Julian, R. K.; Kapp, E.; McComb, M. E.; Oliver, S. G.; Omenn, G.; Paton, N. W.; Simpson, R.; Smith, R.; Taylor, C. F.; Zhu, W. M.; Aebersold, R. Nat. Biotechnol. 2004, 22, 1459. (12) Renard, B. Y.; Kirchner, M.; Monigatti, F.; Ivanov, A. R.; Rappsilber, J.; Winter, D.; Steen, J. A. J.; Hamprecht, F. A.; Steen, H. Proteomics 2009, 9, 4978. (13) Craig, R.; Beavis, R. C. Bioinformatics 2004, 20, 1466. (14) Kim, S.; Gupta, N.; Pevzner, P. A. J. Proteome Res. 2008, 7, 3354. (15) Bandeira, N.; Clauser, K. R.; Pevzner, P. A. Mol. Cell. Proteomics 2007, 6, 1123. (16) Greenberg, R.; Groves, M. L.; Dower, H. J. J. Biol. Chem. 1984, 259, 5132. (17) Boeckmann, B.; Blatter, M. C.; Famiglietti, L.; Hinz, U.; Lane, L.; Roechert, B.; Bairoch, A. CR Biol. 2005, 328, 882. 3939
dx.doi.org/10.1021/ac3034272 | Anal. Chem. 2013, 85, 3933−3939
